Worth Playing? Girls' Frontline: Project Neural Cloud Review

Girls' Frontline: Neural Cloud (also known as Project Neural Cloud) is an upcoming Roguelike strategy game developed by MICA Team/Sunborn, the creators of Girls' Frontline. Readers who have kept up with our previous coverage of this game need no introduction, but for those out of the loop, Project Neural Cloud is a prequel to Girls' Frontline and will be one of the first among Sunborn's four new games to release.

Note: This is a review of Girls' Frontline: Project Neural Cloud as-is from the pre-release version available as part of its Second Closed Beta Test. Many more feature additions, changes, and QoL improvements can be expected between now and official release, just like what we saw with the numerous enhancements made between CBT1 > CBT2.

You don't need to play have played GFL to enjoy this game, though prior knowledge of GFL is a bonus to notice all the cameos and references!

Will I Like This Game?

There are many possible reasons to pick up this game; it's up to your preferences!  

• Interest in the story
• Experiencing the autochess genre
• Character collection
• Visual/audio aesthetics
• Number crunching & theorycrafting

At the core of it all, Project Neural Cloud is a real-time strategy game. There aren't many games like it, so the only sure way for you to know is by trying the game out when it launches. If you want the big picture review, however, read on!

Author's Note

This review will be largely from the perspective of a player who has progressed through most of the available content. I played through the second closed beta (which lasted two weeks from Dec. 3 to Dec. 16) and reached 100% story completion as well as 1560m out of a maximum of 1600m possible in Endless Mode.

Because a lot of the gameplay details have changed from Closed Beta 1 to Closed Beta 2, it's very likely that the exact minutia will be improved again with Open Beta and the eventual official release of this game. If you are familiar with the game in general and want my personal opinions, skip to the Closed Beta Retrospective section. 

Project Neural Cloud is an autochess strategy roguelike mobile gacha game. That's a mouthful!

What does it really mean in the context of this particular game? 

• Auto chess because your units fight by themselves after battle starts without the need (or ability) to manually control them.
• Roguelike refers to the random stage generation that makes every map clear attempt different.
• Gacha in this game so far applies to the unit collection aspect - it's a fairly standard character gacha with per-pull rates similar to other character gacha mobile games. 

Is the game hard?

"Easy to learn, hard to master" is a simple yet fitting description for Project Neural Cloud. 

If your characters are appropriately leveled/geared, clearing fights successfully will mostly depend on your initial unit placement. Here's an example fight from Chapter 4 that shows how re-positioning units after a reset can change the outcome of a lost battle. 

More advanced micromanagement is possible, but not necessary (unless you want to push the extreme end-game ASAP). You can manually use a stun to interrupt some enemy skills, nuke a low-health target, set a mark to burst down an enemy boss, chain synergistic ultimates together, or even abuse AI targeting to split the enemy backline away from their healers/tanks for easy pickings. 

tl;dr: The game does have a lot of strategic depth and a high skillcap; however, while a skilled player might push a little further than a casual player with the same unit lineup, it takes a lot more work. Players who are stuck on content can simply wait and train their units a bit more before proceeding, and they'll have an easy time.

The tutorial and chapters do a decent job of introducing mechanics gradually so that you naturally learn all the tricks just by playing through the main story. As a rough recollection, here's what I remember learning from each chapter released so far in this beta: 

• Chapter 1
  • Function Card Basics
  • Map layouts & destructible obstacles
• Chapter 2
  • How to position your tank
  • Using ranged Dolls (Specialist/Marksman class)
• Chapter 3
  • Focusing down priority enemy threats
  • Enemy Debuffs & Buffs
• Chapter 4
  • Playing around enemy aggro
  • Building around synergistic Function Cards

There's no in-game tutorial explaining any of this - a player will naturally pick it up through re-trying challenging fights until eventually figuring out the specific trick. (You can also brute-force the story by overleveling, but where's the fun in that?)

Doesn't Auto Chess sound like too niche of a genre?

Cleista: I never played auto chess before, and I didn't have any particular difficulty adapting to the mechanics. Compared to that of Girls' Frontline, the user experience in Project Neural Cloud is much better as a new player - everything is intuitive and you don't have to pull out third-party tools to try and teambuild or minmax.

The gameplay of Project Neural Cloud itself definitely has potential for wide appeal, and it's up to the game publisher's marketing department to make PNC a hit.

Xyn: As a Teamfight Tactics player, I can confirm what Cleista says to be true. Learning the game is much easier due to its PVE nature, as the Professor is not actively competing against other players. Items and Synergies are replaced by Function Cards, but at this point it is hard to tell whether that is for the better or worse.

How’s the Meta?

The meta can dramatically shift from minor changes to function cards, and is generally chapter-specific. For example, Fighter Dolls have an easy time in Chapters 1/2, while Chapter 3 is a good chance for specialists to show off, and Chapter 4 is a test of your ranged DPS and healers. 

Unit balance isn't great at the moment, but that's to be expected of a closed beta. I'll go in-depth on this topic in the closed beta retrospective section. 

To the surprise of no one, information about the gacha is the most requested topic from a poll in the Official Project Neural Cloud Discord.

Advanced Searching

This is the main character gacha and costs Advanced Retrieval Instructions to roll.

You can buy more rolls with Quartz Sand (a premium currency) at the ratio of 2000 Quartz Sand per 10-roll. Note that this ratio may change in the future and we do not currently know the monetary cost of purchasing Quartz Sand. 

Character base rarities are as follows:

• R (★)
• SR (★★)
• SSR (★★★)

All units can be raised to a maximum of 5 stars (★★★★★)  with investment. Star rarity dictates unit stats and algorithm space (analogous to equipment slots). This is a reduction of one maximum star from the first Closed Beta, wherein characters could be raised to 6 stars.

Here's an example roll: 

The official gacha rates & pity mechanic (if any exists) are not published, but I got 9 SSR units over the course of about 300 rolls (and most other beta testers got similar numbers), so a 3% SSR rate is probably a good guess for what they went with. Again, this may not be accurate and rates could change in the future. 

Are Rolls Easy to Farm?

You can't farm as many rolls as Girls' Frontline, but you do earn a decent amount just from playing the game. In the Closed Beta, rolls could be earned as follows: 

• 1 Ticket + 50 Quartz Sand per day from Daily Quests
• 1 Ticket + 300 Quartz Sand per week from Weekly Quests
• 30 Tickets per chapter from Endless Mode
• 35 Tickets + >10000 Quartz Sand from Achievements
• >10 Tickets from story quests
• 500 Quartz Sand as maintenance compensation

The one-time rewards are plentiful and there's almost a 10-roll in tickets each week that can be earned from quests, so this is an improvement compared to some of the more popular gacha games out there. 

Can I Spend All My Quartz Sand on Rolls?

Quartz Sand has other uses like purchasing decryption keys (this game's "stamina") or store items, so it is recommended to save Quartz Sand for other uses instead if possible. 

The Mobile Store

This store refreshes twice per day (or you can manually refresh it by spending some Quartz Sand, up to 5 times per reset). Various resources and items for training your dolls can be bought for either Bitcoins (the in-game currency) or Quartz Sand. 

This store is quite useful for both free-to-play players and whales, and it is generally up to the player to decide which items are cost-effective for their current situation. 

Buying Decryption Keys (Stamina)

100 Keys can be bought at a time, up to 5 times per day. The cost increases every time you buy it, starting at 50 Quartz Sand, then 100, 200, 250, and 300.

Free-to-play players are not advised to buy stamina more than once per day, as it becomes less efficient per purchase. 

What About Basic Search?

These are "free rolls" that mostly drop resources, but can yield 1-star and 2-star dolls. Basic Retrieval Instructions can randomly drop from missions and are also generated passively from an oasis building you can upgrade.

At max upgrades (Level 5 Retrieval Terminal), a full Basic Search 10-roll is generated for free every ~24 hours. 

This is going to depend heavily on what you define as "F2P friendly", so I'll instead answer some specific questions that will hopefully give you a good idea. As usual, keep in mind that this is within the context of what is known as of the second closed beta test, and details (especially numbers) may change in the future. 

Do I need to spend money to beat content?

No, but whales may have an advantage in being able to do it slightly faster due to having a more diverse cast.

Do 3-star Dolls powercreep lower-rarity Dolls?

Some 1-star and 2-star Dolls are actually the best units in their own class (at least based on the second beta test's meta), so no.

3-star Dolls do have a slight usability advantage: their ultimate/passive skills are immediately unlocked on acquisition, whereas 1-star and 2-star Dolls will have to be upgraded to 3-star first. 

Do I need to spend money to max out my Dolls (e.g. by rolling dupes)?

There's no dupe-requiring minmaxing system like constellations, potential, or NP level.

Once you roll a certain Doll, you can fully max him/her out over time as a pure F2P. 

Do I need to whale to get most of the characters?

We don't know the exact rates or how many characters will be in the game at launch, so this question (at least when it comes to SSR characters) cannot be accurately answered at this time.

While a large number of free rolls are given, I would expect the developers to adjust rates such that 100% collection will require some spending (since that's their business model after all). 

How much of a progression advantage does a whale have over F2P?

If the F2P player knows what they're doing and is willing to spend Quartz Sand on things other than rolls, they're unlikely to get stuck in main story and won't lag behind in infrastructure. Endless Mode progression may take a bit longer, but can likely still be fully beaten in a reasonable timeframe. (There's no friend support system* in the second closed beta, so this answer is purely from the perspective of your own units.)

In terms of raising dolls, as a very rough estimate, if a F2P player raises their dolls at a rate of "1x", a light spender might be able to manage a rate of 1.2x while a heavy spender would get a rate of 1.3x. Doll rarity is primarily time-gated and it's extremely expensive to speed it up by whaling. 

* If when the game releases there is a friend support system and it's not intentionally crippled, then a F2P player will likely have no trouble progressing just as fast as whales. Borrow a strong DPS and steamroll the enemy. 

How long does it really take to max a character?

Purely counting the neural fragments (puzzle pieces), you need 30 + 80 + 120 + 150 = 380 pieces to reach 5-star rarity. If you only farm on double fragment drop bonus days (every other day), this will take between 4 to 5 weeks depending on how many fragments you need to buy from the shop. 

(Note that because there's a daily limit per character, you'll be farming a few other characters along the way to use up your fragment collection runs each day. This means the realistic rate is closer to 3 characters raised to 5 stars per month based on numbers from the second closed beta, disregarding the Neural Patch cost.)

How bad is the stamina system?

I'll borrow a response from soulmuse to answer this: 

Stamina is reasonably generous. Cap/Regen scale off an upgradable building.

It's hard to say for sure if costs we see now are final (Adv. Algo isn't for sure), but they're not awful. If you fail a mission, you are refunded all but 1 stamina, which is a nice feature which I didn't really expect the game to have.

If you're pure F2P you won't have a lot of stamina issues unless you want to grind 6-8 hours a day or something like that.

If you want to spend your stamina progressing, there's a lot of gameplay hours that can be had from natural stamina regen - you can take your time progressing and you're unlikely to run out of keys.

Alternatively, if you wanted to spend your keys ASAP for some reason (e.g. no time to play), the daily missions/algorithms take less than a minute each and you can empty your whole stamina bar and complete all dailies in under 10 minutes. 

Instant Calculation

If even that isn't fast enough, you can use Instant Calculation Command tickets (5 free/day from a Daily Quest) to perform an "auto-battle" of endless mode for 30 keys and receive all rewards instantly. You can literally empty out 180 stamina in less than 15 seconds!

Fun fact: the instant calculation feature was added during the middle of the second beta after overwhelming demand from players, and the yields from this feature were hotfixed after only 1 day in response to player feedback. The dev team definitely works fast and clearly pays close attention to what the community says - good job! 

There's a dorm system (unreleased), base infrastructure system (the Oasis), and a factory system that is new to the second closed beta. 

The dorm is likely to be entirely cosmetic, the Oasis is just something you upgrade over time for more free stuff like Girls' Frontline's dorm facilities, while the factory system basically generates free energy over time you can use to instantly produce items/resources you need. Here's a demo: 

Factory energy can produce almost everything in the game: Bitcoins, Skill Books, EXP cards, Algorithms, Doll affection gifts, Neural Patches (for upgrading rarity), and much more.

You can assign dolls to each factory to regenerate energy faster (the bonus scales with Doll affection, rarity, and level), but generally speaking the factory will regen from empty to full over a 24-hour day. 5000 energy per day (with 5 maxed factories) is a lot of energy and since unit progress is gated behind factory-producible items, this is an important building to max out, possibly even moreso than the key collection port. 

The art style is, for better or worse, very reminiscent of Arknights. Rumor has it that this is due to the Arknights creator's old tenure at MICA Team, but we'll never know for sure. Character art is generally consistently good, the Live2D + 3D chibis are gorgeous, and players are greeted with breathtaking CGs as they progress through the story. 

For the music and story, you can get a preview in the playlist below (as well as by checking out some gameplay footage). Some readers may recognize this game's OST as the handiwork of Vanguard Sound; they've worked with MICA on many occasions, and their contributions to this game are just as exciting.

The story of Project Neural Cloud covers a lot of material that happens chronologically before the events of Girls' Frontline. The player takes on the role of the Professor who uploads his mind into the 42Lab super cloud terminal "Magrasea" to solve a crisis. It's a standalone story whose first arc seems like it'll last a minimum of 8 chapters (4 were available in CBT2, with the in-game map hinting at a minimum of 4 more sectors being available in the future including the already-teased Pierides and Burbank).

Personally, I quite like where the story is going, and it doesn't look like you'll need any background knowledge of Girls' Frontline to understand it. 

Unfortunately for those interested in the exact story script, you'll need to understand Chinese to read these videos (as they aren't subtitled); translations will be added if they become available.

UI/UX

This is a very interesting topic! My day job is UI/UX (user interface/user experience design). Aside from some minor gripes about buttons being placed in weird places, nitpicks about deep functionality nesting (having to press "back" 4 times to get from Algorithm page back to echelon formation), or interesting functionality like factory pagination not being transparently communicated to the user, I don't actually have too many complaints about what MICA has done with Project Neural Cloud. 

The interface is smooth, modern, and intuitive. Special effects are natural and used sparingly only where appropriate, a consistent theme is enforced across the various views, and the graphical presentation is stellar. EN Players who don't speak a word of Chinese had little trouble navigating through the game screens, which speaks volumes to its intuitive composition. Most things just work and never needed any in-game tutorials to explain. 

Overall I give the UI/UX in this game an 8/10, I'm very satisfied and am looking forward to seeing how MICA improves it even more in the future. (For reference, Girls' Frontline gets about a 4 or 5 out of 10 for UI/UX.)

Note: This section is strictly my personal opinion as a beta tester and will only make sense if you roughly know what the game is about. If you're not familiar with the game in general, read up on the previous sections & articles!

Strategic Depth

The developers really outdid themselves when making the game challenging and keeping it varied. The second closed beta is miles better than the first, with the developers having implemented field objects that have been transformative to the game's overall core playstyle. 

Dying to an enemy artillery spawned on top of a CDR tile? Bait them off of it by placing your entire formation out of range. Boss AoE giving you a hard time? Abuse enemy pathfinding to make it only hit your tank.

Once combined with the chapter-specific function cards that let you amplify the effect of map gimmicks unique to each stage, the game's possibilities are endless!

This beta test enabled a ton of extremely fun builds that I genuinely enjoyed putting together to push content, something that I really wasn't expecting based on what I saw out of the first beta.

When I signed up, I thought I'd be spending most of my time finding bugs and complaining to MICA about game design decisions, but the time I spent on PNC over the past two weeks felt like I was playing an enjoyable and challenging game that I could easily sink many hours into. 

Aren't you overstating this a bit?

Not at all! I think Project Neural Cloud is easily one of the most fascinating strategy games I've ever played, and players willing to go hard at minmaxing will find that there are surprising ways to "solve" certain maps. The core auto chess gameplay works because it's a classic exhibit of the butterfly effect; the outcome of a fight can change dramatically from even simple things like moving one of your Dolls by a single tile at battle start. 

The game has a low barrier of entry, but the potential complexity easily surpasses most other strategy games, even award-winning PC titles. Here's a [translated] quote from a CN player who found out the hard way: 

"You know, here's a funny story: 

Before this round of beta, I reserved a thread on NGA to hold all the guides I was going to write for this game. Then when testing started, I found out I had no idea what I was doing and couldn't beat any of the hard maps, so I had no choice but to ditch..."

Replayability

I had my doubts about how replayable this game would be based on what we saw from the first Closed Beta Test, but I'm very satisfied with the improvements made by the development team before and throughout the second Closed Beta Test. Replayability is excellent and I see plenty of room for the developers to try new things that can improve it even more. 

How much time per day does this game take? 

If you're in a hurry, completing all the dailies, spending all of your stamina, and emptying your factory energy will take less than 10 minutes, and this gives close-to-optimal yields. There's no need to grind the same boring stage for hours on end like Girls' Frontline or certain other games which will not be mentioned here.

The endgame content is pushing deeper depths in Endless Mode, which costs 0 stamina to attempt - so unless you somehow reach 100% completion, there'll always be something to try and beat. 

The production team for Project Neural Cloud have said in previous interviews that they did not want this game to require a lot of time from players, and I think they have done an excellent job at successfully achieving this while still providing an option to play it as much as you want, which I think is an amazing solution that achieves the best of both worlds. You don't have to stop playing if you run out of stamina, and things are still fair for casual players who don't have the time to grind - they won't fall behind. 

The other stamina-based gacha games I've played (Fate/Grand Order, Fire Emblem Heroes, and Arknights) either didn't attempt to solve this problem at all, or provided a lackluster solution that was either limited weekly or tied to events. 

Is there a point to manually clearing Endless instead of using Instant Calculation?

As a matter of fact, yes! While the material drops from Instant Calculation are mostly the same as a manual clear, a manual clear that specifically targets purple (advanced) combat zones can potentially earn gold algorithms (including set pieces).

It's a lot of work and requires a very strong team that can consistently beat those nodes, but if you can pull it off, this is a minmax that experienced players might find worth doing

Character Balance

Characters are... not balanced. There's no other way to put this. Here's Max (1★ base) outdamaging Lam (2★ base) by a factor of 5x in the same fight, despite the statlines being as follows: 

Max: 29% Crit Rate, 162% Crit Damage, 1031 Calculation Power

Lam: 100% Crit Rate, 284% Crit Damage, 1074 Attack

This massive DPS gap is primarily attributable to these factors: 

True Damage

Because enemy HP and Defense both scale up during tougher content, True Damage becomes disproportionately strong relative to conventional DPS in the current game implementation. Not only does it take enemy defense out of the equation, but scaling True Damage usually only requires stacking one stat, instead of having to invest in both Damage and Armor Penetration. 

Unfair Algorithm Space

For some reason, Lam's algorithm grid is 10 Defense/4 Offense/4 Utility compared to Max's 4 Defense/8 Offense/6 Utility. I don't know why the developers did this, because there's no part of Lam's kit that scales off defense.

If her grid were a bit more offense-oriented, Lam could have reached 1.4k Damage while also maintaining much more Armor Penetration, closing the DPS gap significantly. 

Character Scaling

Between the first and second Closed Beta, a new stat, Calculation Power, was added to the game. This stat is now what determines the scaling of character abilities (both active and passive), whereas they previously scaled off a % of the character's Damage, Defense, or Health.

Calculation Power being its own independent stat (considered an "offensive" attribute for algorithms) presents a dilemma: how much can players invest into Calculation Power?

• Medics: They can happily stack Calculation Power and ignore other stats. 
• Defenders: They can stack HP, Defense, and Calculation Power separately. 

What about the other classes?

• Marksman: This is a pure DPS class whose members tend to derive most of their damage from basic attacks. However, their abilities normally scale purely from Calculation Power, and it also doesn't help that many of their Function Cards have Calculation Power scaling. 
  
  This is a problem for most Dolls in this class, because skills usually account for a very small percent of their overall DPS, making it completely not worthwhile to add any Calculation Power at the expense of Damage or Armor Penetration. This can result in interesting outcomes like Lam's DPS decreasing whenever she activates her skill, because a basic attack from her hits for 1k while her active laser beam does 200. 
   
• Fighter: Not only do they suffer the same problem as Marksman-class Dolls, but Fighters also need defensive stats and can't run a glass cannon build. 
   
• Specialist: Thankfully, Specialists don't all scale the same way, and it's often obvious what stat they prefer and they will work well by stacking that particular stat. Antonina, for example, likes Calculation Power, whereas Banxsy prefers scaling her Damage. 
  
  Function Cards for Specialists generally improve their utility, though if Specialists are being used as your main DPS, having one or two Specialists who scale with the Damage stat will help synergize with the DPS-specific Specialist Function Cards. 

To circle back to the above image: why can Max stack Calculation Power? The answer lies within her Passive Skill: each attack from Max applies True Damage equal to 30% of her Calculation Power to up to 2 targets. This is powerful and direct Calculation Power scaling that most Snipers do not have, and lets Max stack a single stat and scale both her DPS and Skill Damage. When most Dolls have their damage scaling off separate, conflicting stats, monolithic skill scaling will come out on top by a huge margin. 

As a side note, skill scaling is the same reason why Betty is the meta warrior in this round of closed beta (her free superarmor is a welcome bonus too). With an active skill that functions as a self-buff that increases Damage and Attack Speed, it's a no-brainer to simply stack Damage on Betty to boost her basic attack. 

The difference is so great, a 3.5★ Betty can outdamage a 5★ Sol despite Sol having True Damage on her attacks. 

Will this be fixed? 

Fortunately, the developers seem to be aware of this problem (as they address it explicitly in their CN feedback responses) and know how to fix it. Looking at the difference between CBT1 and CBT2, a lot of minor changes have pushed around the unit power levels to much more reasonable levels. The main problem offenders for this beta, Max and Betty, were not playable in CBT1, so understandably they would not be as balanced as characters who have already received a round of testing. 

Minor game mechanic changes can also significantly change a unit's viability (for example, if Max's true damage can no longer crit, her DPS would literally drop by 9-10x), so it's very likely that the power dynamic will shift tremendously in the next beta for this game. 

Is this game worth spending money on?

I whale in almost every gacha game I play, so I may not be the best person to ask this question to. I do like Project Neural Cloud a lot and can see myself spending money on it, or at a very minimum buying the monthly card (if there is one).  

That being said, if we assume 2000 Quartz Sand (a 10-roll) costs $20 (a reasonable guess based on other character gachas), I probably wouldn't buy rolls with it if nothing changes about the game's monetization; I'll likely still buy some premium currency (on top of any monthly card that is offered), but I'll opt to purchase stamina or shop items with instead. This is due to a variety of reasons: 

Lack of value for unit dupes

Don't get me wrong, I absolutely abhor gacha mechanics that force you to roll the same character multiple times. However, the other extreme like what MICA did for this game isn't really ideal either. Consider the following gacha roll: 

I received one new unit, Zion, which is great. However, consider all the 1-star rolls: because I already owned all of these units, they got auto-converted to materials that are practically worthless in the amounts given. Even though I rolled Abigail 3 times, I'm no closer to raising her to a higher rarity with this roll than if I rolled her 0 times, which feels bad as a player and discourages me from wanting to spend more money. 

Due to how raising unit rarity works in Project Neural Cloud, it costs the same number of Neural Fragments (puzzle pieces) to raise a 3-star unit to 4 stars, regardless of the base rarity. However, because 1-star and 2-star units have to be raised to 3-stars with even more fragments, and the cost of each fragment scales up linearly based on how many you have bought already, it's actually far more expensive to raise a low-rarity unit than someone who starts at 3-star, which runs counter to how this normally works in other gacha games. 

Suggested solution: In addition to the raw materials given here, give one or two neural fragments (puzzle pieces) for the characters you roll. It won't affect progression much (unless you spend megawhale amounts of money), but it'll help close the cost gap between low- and high-rarity units as well as make gacha dupes feel more meaningful to players. 

Stamina Cost

Buying stamina once per day is pretty cheap, but buying it more times gets exponentially more expensive fast and is the only way someone can overcome the progress timegating in this game.

With an assumed cost of $20 per 2000 Quartz Sand, buying 100 stamina costs 50 cents while buying the maximum 500 stamina every day costs $9, a considerable expense for whales trying to push as much content as possible ASAP. Since I'll probably want to push new content ASAP, I don't want to spend my Quartz Sand on rolls if I can buy stamina with it instead. 

The Shop

I don't particularly recommend refreshing the shop (unless you're a whale trying to leave early game hell ASAP), but if you do end up refreshing the shop its max 10 times a day, that's another 500-1500 Quartz Sand gone based on how lucky you were with the item rolls and how much you buy, 

Personal Preference

This is probably the most controversial point in here, but since I do not care too much about hitting 100% collection, I don't feel strongly about a need to buy gacha rolls if I don't have to; after all, as long as I can experience all the content and there's a viable way forward (meaning I'm not hardgated by missing a certain character), I'm happy to stick with a limited roster. 

The good news for F2P players is that Project Neural Cloud so far doesn't look like the type of game where your progress is gimped by not having a specific character. While a lot of beta testers have felt this way due to Persica having a hard time solo-healing without Gin, I don't think this will be a problem in future iterations of the game; more healers will probably be added, and Persica is actually a very strong healer once you level her active skill to Lv. 4, easily capable of solo healing through Chapters 3 and 4. 

Miscellaneous Complaints

The game isn't perfect by any means; this section will be dedicated to miscellaneous specific problems I want to rant about that I'm pretty sure the developers know already (since I've seen plenty of CN testers complain about the same issues). 

The "Circuit Breaker" function card is broken. In a bad way. 

When a Guardian receives more damage than 10(-1)% of their Max HP, the damage will be capped to 10(-1)% of their Max HP. Minimum of 5%. 

On paper, this isn't such a bad card: when bosses hit you for 5 digits worth of damage, tanks absolutely need it to survive. The problem is that this card can be used on a level 1 Guardian, so you can turn a squishy unleveled unit into an unkillable juggernaut and use this to cheese content that shouldn't otherwise be beatable. 

Suggested solution: Add an internal cooldown to this mechanic, which still lets the card do its job of negating infrequent big hits without making it abusable. 

Algorithms are too hard to optimize

Algorithms are basically the "equipment" of this game. They have a random base stat, random bonus stat lines, and random tiers of bonuses (for instance, a +Damage line can be +3, +6, +9, +12, or +16 Damage). Plus, they can be from a random set that has nothing to do with the stat that's on the algorithm. Rolled a +Damage set piece but the base stat is Calculation Power? It's now vendor trash. 

It's possible to enhance these bonus lines to up to 5x their base value, but doing so requires a material that can only be acquired by dismantling orange (ultra rare) algorithms. Rerolling the lines is even worse, requiring both a material crafted from disassembled algorithm parts and an identical algorithm (in both size, area, and set). 

This closed beta test included "advanced algorithm" farm stages specifically for testing purposes that had 20-30 times the normal algorithms drop rate, and it still took thousands of stamina and hundreds of thousands of bitcoins to craft the following mid-tier algorithm set piece. 

If the drop rate and crafting mechanics aren't reworked or otherwise improved, something like this could easily take years to craft in the live game, and that's not something I think any player wants. 

While I love the basic idea of algorithms (reminds me of affix crafting in Path of Exile), re-crafting algorithms here is too expensive and the stat bonuses are far less impactful. This system needs a lot of adjustment; some of my ideas are as follows, and hopefully the developers can make the necessary adjustments to deliver a better experience when the game launches for real. 

Suggested solutions: 

1. Make algorithm rerolling cheaper or more deterministic.
   
   Right now, set pieces are already hard to come by, and you need to sacrifice a set piece to reroll its lines completely at random. With so many affixes and tiers available, the chance of getting what you want is virtually nil - you could end up sacrificing hundreds of set algorithms (not even a quantity any normal player would see in real gameplay) to craft a perfect one. 
   
   The following changes should massively improve the player experience without affecting balance too much: 
    
   • Random rerolls should not consume an identical algorithm. It's random, and the rerolling currency already costs disassembled algorithm parts - this is gated sufficiently already. 
      
   • If the player does sacrifice an identical algorithm, this should be a deterministic reroll - i.e. affixes from the fodder algorithm are transferred to the one being rerolled. 
      
2. Allow the base stat to be rerolled. 
   
   This removes the problem of set effects completely mismatching the base stat on the algorithm piece. 
    
3. Allow enhancement of blue and purple algorithms at a significantly reduced cost.
   
   Lower rarity algorithms are not currently competitive, and gold algorithms are too expensive to enhance. Letting low-tier algorithms be enhanced serves as a nice middle ground to let players trade some long-term progress for an immediate stat boost, and adds more variety to the possible minmaxing that can be done on the algorithm board. 

Need for better camera control

While the default top-down strategic view is useful for easy player control in challenging content, the camera controls are somewhat lacking. While it's possible to rotate and zoom in, there's no option to "follow" a character via third-person view or easily get a close-up of the action, which is a shame - the 3D chibi models are pretty cute.

Suggested features: 

• An option for the camera to follow a specific character
• Possibly a setting to tone down SFX unrelated to the followed character
  • For instance, if you are following the up-close melee action for Sol, the game shouldn't blind your screen by rendering the hacking SFX from your Antonina or the nuke from your Hubble
• Maybe an option to auto-skip ultimate animations after it's been skipped once in the current map - the 3D SFX is pretty good-looking too and sometimes I want to see that in action without that split-second of 2D cutin that I'm going to skip anyway. 

As a little bit of trivia, Imoko (MICA's in-house artist who drew G36 and QBU-88 among a few other dolls and a lot of promo images/login backgrounds) seems to be the designer for Project Neural Cloud's 3D chibis, according to her weibo post. 

Not much to do other than wait if a player is stuck on content

Like all mobile character collection games, newly-acquired characters have to be raised and equipped to some degree before they're really usable in battle. For Project Neural Cloud in particular, this means raising a character's rarity, level, skills, and equipping them with algorithms that boost the stats they need. 

• Level is easy to raise - feed some EXP cards, and you can get a character up to speed in seconds. 
• Algorithms must be farmed over time, which is OK - the player will naturally stock up on algorithms even if nothing is wrong. 

The two main problems here are rarity and skill level.

Rarity is self-explanatory: it's one of three major determining factors behind your character's stats (the others being Level and Algorithm loadout). Not only do you receive a massive boost to stats just from increasing rarity, but higher rarities increase the amount of algorithm board space available, allowing even more stat growth. 

Compare the below stats and look at the massive difference between 3.5 stars and 4.5 stars (a similar gap exists between 2.5 stars and 3.5 stars)

Unlike the other two, however, it's much harder to stock up on materials to raise a unit's rarity - you're timegated on a few days of fragment daily farm, unless you're willing to pay the exorbitant shop price for buying fragments instantly. 

This creates an problem wherein if a player decides they want to play a new character, they are forced to either whale a lot or wait a week or two to get them up to 3 stars - which isn't a great experience. 

Skill levels require skill books to raise, which are expensive; as an important note, characters often hit a massive powerspike when their skill is raised to Lv. 4. For example, here's Persicaria's heal: 

Level 3: Heals the lowest HP target, the heal can bounce to 1 additional ally. 

Level 4: Heals the lowest HP target, the heal can bounce to 2 additional allies, and healed allies gain a shield that negates the next 2 attacks.

The difference is massive and a lot of characters have similar amounts of crazy skill scaling on at least one of their skills - so any new characters you roll from the gacha will not feel very strong until you train their rarity and skills up, which might take a few days. 

Suggested solutions: This is a tricky problem to solve, because it's both important to have long-term character growth be a feature and ensure a new character's immediately usable for players who want to try them out. I have a couple of ideas that might work: 

• Stat Equalization: When playing through a map that has been cleared already, allow the player to turn on this option to temporarily buff the weakest Doll in their team as follows: 
   
  • Their rarity is treated as 0.5 stars below the team's median rarity
  • Their level is treated as 90% of the team's median level 
     
• For example, if you want to try out a new 1-star unit and you put her in a team with Level 60, 3.5-star Dolls, then the Stat Equalization feature will make it so that your Level 1, 1-star unit is treated as if it's Level 54 with 3-star rarity in terms of base stat calculations. 
   
• A natural way to implement such a feature would be adding it as an overclock buff that can be selected with a bitcoin cost. This will let players easily "preview" their units at a cost, without disincentivizing actually leveling them. 

Aside from the above solution, some other more general recommendations could also work: 

• Adjusting main story difficulty so that the normal difficulty stages can be cleared with 3-star units only, with Hard (Nightmare) and Endless modes balanced appropriately against higher-rarity units. 
  
  This can remove the frustration for players who are stuck on story while still incentivizing them to train their units to beat Nightmare and Endless stages. 
   
• Re-balance endless mode scaling so that the difficulty ramps up smoothly like a smooth line, rather than the current situation where Endless difficulty feels more like a staircase with massive jumps in difficulty every 40 meters (4 stages).
  
  The current (CBT2) implementation of Endless Mode is such that players often end up in a case where they either steamroll the content or are steamrolled by the content - very rarely will they find their currently-challenged depth to be of similar strength as their team. 

Ultimates are not balanced

There's a joke among round 2 Beta Testers that Hubble's smile is burned into their retinas: because her ultimate is just way too good compared to that of other characters. Even after the developers nerfed its duration to half of what it used to be in the first Closed Beta, it's still miles better than every other ultimate. 

In general, ultimate skills that don't apply some kind of additional utility almost never see use in Project Neural Cloud; the new Calculation Power stat is, of course, partly to blame for that. 

Here's my personal evaluation of every ultimate that was available in the second round of Closed Beta, in descending order of usefulness: 

Always worth using

• Gin: It heals your characters and boosts their Damage and Calculation Power, which both means more DPS and more healing (while it lasts). Pretty much useful in almost all situations. 
   
• Hubble: It deals true damage, applies a 2-second stun, and it bunches up enemies to get hit by your AoE attacks. What's not to love about it?

Great but situational

• Antonina: It disables all enemies for a decent amount of time, which can be good to buy more time to DPS down an enemy before its next skill activation. 
   
• Persicaria: It's a heal that also reduces damage taken for 4 seconds, which can be worth using over Gin's heal if there's an incoming attack that would one-shot you otherwise. 
   
• Jessie: This is a unique ultimate that purges all allied debuffs and enemy buffs, which can be incredibly strong in certain fights such as the Beelneith or Mara boss battles. 
   
• Sol: She has one of the rare few damaging ultimates that still have scaling based on the Damage stat. As a result, a well-timed Sol ultimate can one-shot bosses if she's built appropriately, giving this some use. 
   
• Fresnel: While the damage output is lackluster, the [Blindness] debuff (80% chance to miss) applied by this ultimate is a huge short-term survivability boost that can be used to great effect. Plus, the animation looks amazing.  
   
• Simo: This ultimate cloaks your units for a few seconds and heals them, which will occasionally see use in some particularly tough fights. 

Possibly can see use

• Chelsea: This is a high damage ultimate that also happens to be on a character who can comfortably stack Calculation Power. 
   
• Croque: The damage output is bad, but if your Croque has a lot of Calculation Power, this can be used to instantly destroy roadblocks on round start - which has some strategic value. 

Worthless

• Lam: Calculation Power scaling on a character who prefers stacking Damage. 
   
  • Suggested fix: Add Damage scaling or significantly increase its multiplier. 
     
• Betty: Poor damage output plus Calculation Power scaling on a character who prefers stacking Damage.
   
  • Suggested fix: Add Damage scaling, increase the bleed stacks, or significantly increase its multiplier. 

EXP Mission drops are too RNG

This is a small thing but it is so strange that I have to call it out explicitly. I don't know if they fixed it by the end of beta or if I just got lucky (since I haven't gotten the minimum drop again in the last few days), but these two drops are both from a clear of EXP Episode 6 - the highest possible tier of the EXP daily. 

This is a 36.25x difference in the amount of EXP that can drop from the same mission, which is ridiculous. Some variety is fine and might even be desirable (such as dropping between 4-6 orange EXP cards), but the fact that it can drop zero shouldn't happen, and I hope the developers fix this. 

Endgame algorithm farming

This is certainly better than the first Closed Beta Test which had no real endgame to speak of, but because there's a strong timegate on everything else (farming fragment pieces, factory energy, etc.) the stamina expenditure endgame after beating all of the story basically boils down to two things: 

• Farming raw materials for the factory in Endless Mode
• Farming the algorithm missions

I don't personally have a problem with this, but the algorithm system received mixed reviews from CN players, so it's likely that there'll be some changes made to it for the next rounds of beta. I think the system has potential to be really good, so I wonder what the developers will do about it. 

I'll use this section to list down some of my miscellaneous observations during this Beta Test as well as put down some wishlist items that don't fit into any other section.

I don't know if the developers will ever read this, but if anyone from the team does happen upon this article: thank you for doing a great job on the game so far. I'm sure the rapid game patches and development turnaround time is a result of many sleepless nights and I wanted to let you know that the effort to respond to community feedback is well-appreciated even by western fans.

Voice Acting & Character Interaction

Voice acting and character-specific dialogue were not added to this beta test, since this test appears to still be focused on iterating the core gameplay loop.

If they get added in the next beta, I'm pretty interested in seeing MICA's casting choices for the original characters in this game, as well as the dialogue for any Dolls who will show up in Girls' Frontline later. 

Some particular things I'd really like to see: 

Specific responses when gifting items

Every Doll has gift items that they like, feel neutral towards, or dislike (for example, Coffee gives more affection when gifted to Persica). Currently, gifting these items doesn't do anything except increase a number, and I think it would be a really nice touch if we could get some character-specific lines that are said in response to these types of gifts. 

In-game 3D Model Viewer

The game's 3D chibi models are amazing, but it's really hard to see them in action especially with the awkward camera controls and so much flashy SFX during battles. Sometimes a serendipitous shot reminds me of how good they are and I lament the lack of a way to see all these animations/expressions within the client.

Function Card Balance

Most function cards have interesting mechanics and while there are obviously strong and weak cards, this balance can be improved just through numbers-tweaking. Since I'm not sure what other gameplay changes will happen in the meantime, it might be premature to give fine-tuning suggestions that may be thrown out of whack by a core mechanic change, so I'll skip critiquing most of the cards that just need a number adjustment. 

One thing that does need more clarity is the fact that most players were unaware of the fact that a maxed-out Overclock Tower lets you start a map with any one function card of your choice. This is a gamechanging buff that can let players push way further in content than otherwise possible, and save a lot of headaches over savescumming for a particular card. 

The simple solution to this would be showcasing every possible overclock during the Overclock Tower tutorial, instead of forcing the player into selecting a relatively negligible 5% max HP boost. 

Endless Mode isn't truly Endless

The "Endless" mode caps out at 400 meters, and it's definitely a downer to realize you can't go any deeper (I managed to hit the cap on all chapters except Chapter 3 during the 2-week beta test).

I hope MICA figures out a way to make it truly endless for players who want to test the limits of their teams against the hardest possible content. 

Some regular enemies are stronger than bosses

Most regular enemies have interesting skills that make them a lot more threatening than just a sack of stats.

The generic ranged enemy deals more and more damage based on your % missing health, the generic enemy healer buffs enemy Damage and Calculation Power per heal, and the wolf assassin enemy ("Patience") gains additional Dodge and Damage every time an ally dies. 

Because some enemies grow exponentially more powerful if you don't take them out first, this can lead to a very interesting outcome where your team has no problem dealing with the chapter boss but gets wiped out by a single enemy mob.

This particular mechanic is intended to be played around by simply focusing down any assassins first, but there has been significant confusion and complaint about it being too strong from the CN beta testers. Whether that's the player's fault for not reading or the developer's fault for making content too difficult for casual players... I'll leave that decision up to the reader. 

Hopefully, the developers can address this in a way that appeals to the casual audience without making the game too easy for players who want a challenge. 

No in-game explanation of damage calculation

Is 75 Damage better than 35 Armor Penetration, or vice versa? This question isn't answered by the game anywhere except through its "combat effectiveness" calculation, which is unfortunately a pointless value just like it is in Girls' Frontline. Similarly, the "attack speed" stat makes no sense - doubling your Attack Speed does not actually double your DPS or halve your attack interval. 

What's the exact math behind these stats? Hopefully the developers can make this clearer within the game itself. 

Random things I noticed during the Beta

• The art assets for this game are in 2048x2048 - four times the pixel resolution of the Girls' Frontline in-game character art. 
   
• The story script was significantly altered between CBT1 and CBT2 (and the Nightmare difficulty side stories got cut), so it is unknown whether or not it'll be further altered for future versions of the game. I'm hoping the cut content will come back - the dialogue was pretty cute. 
   
• Based on what I've seen on CN social media, there's an inverse correlation between how easy someone says the game is and how much of the story they have cleared. Curious. 
   
• For players who fail a stage halfway through, they are given two options:
  • Spend partial stamina (based on current clear progress) and receive partial drops. 
  • Alternatively, they can refund all but 1 stamina, but forfeit their drops. 
     
• Maintenance Mail only lasts 4 hours and happens during the middle of the night (for the US), so I was never able to read the maintenance changelogs unless someone else took a screenshot. This is probably only a problem for the Closed Beta though. 
   
• It's possible to minmax characters creatively and make some pretty fun meme builds. Machine gun Simo is one of the most hilarious examples of this that I've seen. (source)

FAQ

Is the second closed beta representative of the final game?

Well, the beta client does have a lot of watermarks to remind people "NOT REPRESENTATIVE OF FINAL PRODUCT" everywhere. 

On a serious note, the core game loop looks to be fairly established, but a lot of the QoL bells and whistles have yet to be added. For example, voice acting, additional Live2D (Simo got a Live2D this time and didn't have it in the first CBT, for instance), the unit index and 3D Dorm system, and so on are not revealed yet.  

Most of the WIP features are strict improvements, so regardless of what you think of this beta, the final release is very likely to be even better. 

I'm not very good at strategy games, will guides work for an RNG Roguelike?

Generally speaking, a good synergistic team composition will work across many different runs even with vastly different function card loadouts. The three backup unit slots can address this problem: if you field Fighters but mostly get mostly Marksman cards, you can swap out your DPS and keep going without having to reset. 

Plus, most maps are won by smart unit positioning (which is something that stays the same between different players trying to clear the same map, since the map layout is consistent). When I couldn't beat Algorithm Collection E6, I consulted a CN clear video and used the same layout (albeit with slightly different dolls), which worked out for me. 

Can my phone run this? It looks pretty graphically intensive. 

The official recommended specs are "Snapdragon 670 / Snapdragon 820 / Kirin 960 / HelioX30 or better processor", but the beta appears to run fine on lower spec phones without much issue. Turning down the graphical settings can help performance as well. 

Based on my experience running this on my phone, the PNC beta client's performance is actually much better than what I get with the Girls' Frontline EN client. My phone runs a decent amount hotter though (45°C running Project Neural Cloud vs. 32°C running Girls' Frontline). 

Was the beta very buggy?

I hard crashed a few times on my emulator (MuMu), but the client was stable from my phone. Most of the "bugs" I noticed were more related to game balance and not the game engine or usability. 

When will the global version come out?

There's no announcement even for the CN release date yet, so we don't know. Hopefully, it won't be delayed much. 

Can you compare PNC to [insert game here]?

I try to not make these comparisons often because it's a very different game from what's currently on the market. What I can say is that the production value is above average (though obviously not AAA level) and the game's user experience has avoided most of the mistakes that haunt older titles.  

Have the game's calculation formulas been datamined? 

Yes, but these things can easily change in the future. Function card math changed significantly between CBT1 and CBT2, for example.

Is there a tier list of the Dolls available? 

Like I said, the game will very likely change so don't get too ah-

Please post chapter clear guides!

...

Afterword

Note: most of these articles have not yet been updated with CBT2 info. Please wait patiently!