DEVELOPERS: What are some of your bottlenecks, and how did you widen them?

[CurvyLinesEverywhere]

We all know that the single most important thing you can do to make your game shoot to the top of the Patreon charts is to release more updates. But how do you do that? Surely every dev would update every month if we could, but what's holding us back? What are the

You must be registered to see the links

in your update pipeline?

It's worth considering this question in terms of one-time processes VS recurring processes. For example:

Workflows
(Per-project, usually developer-specific)
- Use a real-time character posing tool such as Honey Select or Koikatsu instead of Daz Studio or Blender's slower pre-rendering.
- Unity Workflow for developing new game mechanics: ( Backend > Features > Content )

One-time Processes
(per-Project or per-Character/Scene)
Concept Design (Deciding what to make)
-

You must be registered to see the links

(How to prioritize & make good decisions when you have no data to draw from.)
-

You must be registered to see the links

(Set a Time & Place to think creatively, use of Humor, Closed and Open modes of thinking.
Character/Clothing Design
Environment Design
Materials Conversion/Customization
Model Conversion to/from Daz Studio
Training/Learning to use the various Software
(Though, realistically, this is recurring, there are often diminishing returns past a certain point where basic daily use becomes "easy.")

Recurring Processes
(per-Render or per-Line)
Scene Composition (Director decisions, Scene loading, Camera position and orientation.)
-

You must be registered to see the links

(Correctly!)
-

You must be registered to see the links

-

You must be registered to see the links

-

You must be registered to see the links

Character Loading and Posing
-

You must be registered to see the links

-

You must be registered to see the links

(Requires an extensive Pose Library, or combine with the next tip.)
-

You must be registered to see the links

(Because Blender's custom posing tools are SO much faster and more intuitive.)
Animation
Rendering
-

You must be registered to see the links

Writing
Gameplay Programming (Choices, mini-games)
Playtesting
Releases
Storefront Maintinence (including Steam/Patreon/Various tip jars)
Promotion/Marketing Funnel
Backer Rewards

Fast, Cheap Hacks
- Fix it in Post
- Fake It If You Can't Make It (Or sometimes, even if you can.)

For me, my single most time-intensive recurring process used to be Render Times, until I learned

You must be registered to see the links

and use the (basically realtime) Eevee renderer. Right now, I suspect it's probably character posing and scene composition. IT Roy had a great tutorial about

You must be registered to see the links

, but it depends on a robust Runtime full of poses and, you know, actually using Daz Studio. While I could pose in Daz and export the whole scene to Blender, that would prevent me from using custom materials unless I wanted to manually re-apply them after each import. And custom materials are basically essential to getting Eevee renders looking anywhere near good.

What about you? What's the slowest part of your process? And how did you speed up a step that used to be your bottleneck, but isn't anymore?

 

[MissFortune]

I'm at a point with Daz where I can realistically spend a bit of time and come up with (or something close to) what I've envisioned in my head. So, that makes Character creation and setting up environments fairly quick. Being advanced in Photoshop also helps speed up material customization a fair bit. I've been trying to find time to sit down with Blender, and at the very least, learn the transfer and animation process. I will eventually when I find more time to do so (likely after I graduate.). I've been playing around a bit with Substance Painter and Daz Figures/materials, though, and man do

You must be registered to see the links

ever make things easier.

I thankfully have a pretty wide array of experience in quite a few artistic mediums, either via professionally or just from the environment I grew up in. So, stuff like composition and angles, etc. come pretty easy. Same goes for writing. That's why I try to help where I can/am able to. Many don't have or didn't have an ability to pick up the things I was lucky enough to.

But, in short, I'd say my bottlenecks are social media/promotion and rendering. The former being that I'm not exactly an extrovert (though, I'm trying to be. Self-improvement's a bitch, eh?) and just trying to figure out that 'right way' of doing things. The latter being more speed-based. I'm on a 3080 (which is definitely adequate, obviously.), but I'm usually waiting for renders to finish longer than I'm working on them (which is more of a literal bottleneck, I guess?). Kinda regret not grabbing a 3090 when I was able to lol.

 

[CurvyLinesEverywhere]

I'm at a point with Daz where I can realistically spend a bit of time and come up with (or something close to) what I've envisioned in my head. So, that makes Character creation and setting up environments fairly quick. Being advanced in Photoshop also helps speed up material customization a fair bit. I've been trying to find time to sit down with Blender, and at the very least, learn the transfer and animation process. I will eventually when I find more time to do so (likely after I graduate.). I've been playing around a bit with Substance Painter and Daz Figures/materials, though, and man do

You must be registered to see the links

ever make things easier.

I thankfully have a pretty wide array of experience in quite a few artistic mediums, either via professionally or just from the environment I grew up in. So, stuff like composition and angles, etc. come pretty easy. Same goes for writing. That's why I try to help where I can/am able to. Many don't have or didn't have an ability to pick up the things I was lucky enough to.

But, in short, I'd say my bottlenecks are social media/promotion and rendering. The former being that I'm not exactly an extrovert (though, I'm trying to be. Self-improvement's a bitch, eh?) and just trying to figure out that 'right way' of doing things. The latter being more speed-based. I'm on a 3080 (which is definitely adequate, obviously.), but I'm usually waiting for renders to finish longer than I'm working on them (which is more of a literal bottleneck, I guess?). Kinda regret not grabbing a 3090 when I was able to lol.

Since it sounds like you have an actual budget, have you considered

You must be registered to see the links

Why update your graphics card when you could rent uptime on a thousand of somebody else's graphics cards?

 

[CurvyLinesEverywhere]

My biggest time sink was an improper workflow, I think.
It took a while to discern that developing my projects in a specific way yields the best results.

For me it looks like this, backend improvements > new features / implementation > feature testing, bug tests > new events / content > final playtest.

I found doing it in any other order greatly inflated development time.
If I had to identify any part of the process within this workflow that could improve overall - hmm, new features and testing is the most time consuming. It easily accounts for 50% of my development cycle.

In an ideal world I could have some dedicated play testers to help cut down the amount of time dramatically. But as of now, I'll probably just hire someone dedicated for the task.

In the past, creating art was the biggest issue but I use various programs and services to circumvent that problem.
View attachment 2120852
This, for example was made using Canva and some model rendering.

Are those banchouforte's models with tweaked materials? I have to say, for what they are, you've got them looking fairly polished.

I've literally never heard of Canva before, and when I google it, it looks like some kind of wordpress theme or something. Can you explain how you integrate Canva into your process? What's it for? This screenshot was clearly made using Ren'Py, if the menus are anything to go by.

 

[MidnightArrow]

The biggest roadblock is using Daz Studio. The posing sucks, the animation sucks, the viewport navigation sucks. The common defense people give is "you need to start from a premade pose and tweak it."

No you fucking don't.

Export to Blender and you can make the pose you want in less time than it takes a Genesis 8 to load.

 

[MidnightArrow]

To be fair, I don't use Daz. I find it needlessly complicated and way too time consuming.

I prefer Honey Select 2 for realistic models, but I like the anime aesthetic.
It is, by far, the most used software for character rendering though. (At least in this community).

I've never used Honey Select or Koikatsu, but if either of them have a good FK/IK switch rig, they're automatically more advanced than Daz Studio.

 

[CurvyLinesEverywhere]

Do you always go around accusing people of stealing someone else's work? Odd way to start a conversation.
Edit # 1 :
To answer your question though, no these are models made using Koikatsu / Charas Studio. Many people on this forum use 3D rendering programs to create characters / models for their games.

In general, Honey Select, Daz Studio and Koikatsu are the 3 most used, with Daz being the #1 favorite by developers (here at least).

Edit# 2:
I use canva to create user interface elements like buttons, gifs, videos, backgrounds for screens or pretty much anything else I want.

I think I'll just leave it at that, I'm not too sure I like the way you're speaking to me.

For the record, I wasn't accusing you of anything. My apologies. Banchouforte

You must be registered to see the links

, which they say are made without referencing any other existing model or image, which under US law, makes them legally usable to make new models. If the character isn't copyrighted and the mesh was released waiving its copyright, I don't see where a legal challenge can come from.

In fact, it's vaguely plausible that Koikatsu or Charas Studio could have released a model derived from some of banchouforte's work that you ended up using, and both they and you would have been in the clear to do so.

(I'm not a lawyer, though, so feel free to correct me if you think I'm mistaken about the legality of making derivative works from banchouforte's models.)

The only reason I brought it up was because I'm seeing some superficial similarities, especially in the hair. Although, upon closer inspection, the fingernails and some of the face details seem a lot cleaner than banchouforte's, especially his earlier work.

 

[CurvyLinesEverywhere]

The biggest roadblock is using Daz Studio. The posing sucks, the animation sucks, the viewport navigation sucks. The common defense people give is "you need to start from a premade pose and tweak it."

No you fucking don't.

Export to Blender and you can make the pose you want in less time than it takes a Genesis 8 to load.

Have you made a game this way yet, or is this all still theoretical?

I can confirm that Blender's posing and animation tools are much nicer than Daz's, especially once you rename the bones properly and add some IK and such. I'm unsure if setting up > 80 renders per day would be faster or slower than IT Roy's workflow. It would probably result in your game having a more distinct and unified look, since all the poses are made by the same person.

I'd love to hear more about your pipeline, in general.

Edit: I checked out your patreon, and found out about how you

You must be registered to see the links

and then

You must be registered to see the links

your project,

You must be registered to see the links

. Good luck with it. Disabling joint rotation locks in both programs did the trick, huh? Fascinating.

I'm still actually curious whether you import the posed characters or do the pose and the character separately. In the past, I've found that importing a Daz scene with posed characters results in Blender treating the new pose as that figure's default pose. The slightly clunky workaround is to import poses manually using the Diffeomorphic tool, or just pose absolutely everything manually from the ground up in Blender.

It certainly sounds promising. If I could settle on a premise for a project, I'd start bashing away at it, myself. The goal is to optimize for speed, of course, but if this works, it would also lower the cost of acquiring resources, albiet at the expense of greater complexity in the "One Time" (per character) tasks.

 

[CurvyLinesEverywhere]

Koikatsu Made Character Examples:

• You must be registered to see the links
• You must be registered to see the links
• You must be registered to see the links
• You must be registered to see the links

I linked this because the artist you mentioned seems fond of re-creating popular anime characters in their style. There are places on the web filled with high quality, well done Koikatsu cards of various characters. Though, most creators keep the cards to themselves.

This is also another one I recently made a few weeks ago in about an hour (pardon the nudity)
View attachment 2121293

Those actually look a LOT smoother and more hand-drawn than anything banchouforte's released so far. If you could export a model from Koikatsu and compare it to one of theirs, it'll probably show some obvious differences in topology in Mesh Edit mode.

EDIT: Upon closer inspection, the Koikatsu models appear to have

You must be registered to see the links

than banchouforte's works. (Although the unreleased Naruto characters on his twitter look quite promising.)

Another, more recent example is Vroid Studio, which has absolutely fantastic facial morphology, but unfortunately it only outputs triangles, so you can't subsurf it.

By contrast , Honey Select is made by the same company (Illusion soft), the difference is the base model, which look more like this:
Via Eternum:
View attachment 2121310

With the right filters and lighting, you can make some really nice realistic stuff. In either case, they both run on the same engines for the most part (Koikatsu and HS).

Honey Select I've seen before. It walks this weird line between realistic and stylized. As usual with Japanese meshes, its biggest strength is the face, and its biggest weakness is the lack of hand/foot detail.

The added benefit of both programs is that you don't really have to spend hours creating renders. Aside from posing characters or props, I just need to configure the output size once and press a button to capture a render.

I wonder how big the time savings actually is once you account for scoring the web for Daz/HS content to use, though. Or making your own. DS definitely has a terrible UX though.

 

[MidnightArrow]

I'm still actually curious whether you import the posed characters or do the pose and the character separately. In the past, I've found that importing a Daz scene with posed characters results in Blender treating the new pose as that figure's default pose.

I need to teach my dev partner how to do this anyway, so I think I'll just make a tutorial about it later today.

To answer your question, I export in an A-pose because (as you said) exporting a character already posed makes the pose the rest position.

 

[Saki_Sliz]

My turn!

Most of the time my bottle neck is project momentum, I haven't really released anything, and I usually don't like to work alone, but its hard to keep other teammates interested when I only get the weekends to work on the project, and most of the time I'm burned out from normal life. I've worked on a lot of things to speed up my workflow, from making drag and drop modular code in unity for common game mechanics, to currently developing my own syntax language that acts as both project notes, and can be interpreted into executable code for projects, also so I can get working demos faster...

but now I'm running into the issue where art assets is the main limitation. I want lots of art for interactive scenes and unless its a project an artist is passionate about, I know it can be challenging to get a consistent source of art, hence the past 8 years of trying to learn art and trynig to figure out a work flow. While I have started to hone down what it is I want and can do with my art, its still the simple issue of not being able to spend enough time on a project each month. Which is why I can see how it is important to be able to work on a project full time rather than with just free time... which I'm conflicted about since I finally started my career as an engineer and really enjoy the job and I kind of need a stable income for the next few years since by the end of the next summer I'll have to make some major changes to life (need to move and buy my first place).

The one thing that really help is having enough money to commission writing to help progress the project's story while I focus on technical stuff.

 

[CurvyLinesEverywhere]

My turn!

Most of the time my bottle neck is project momentum, I haven't really released anything, and I usually don't like to work alone, but its hard to keep other teammates interested when I only get the weekends to work on the project, and most of the time I'm burned out from normal life. I've worked on a lot of things to speed up my workflow, from making drag and drop modular code in unity for common game mechanics, to currently developing my own syntax language that acts as both project notes, and can be interpreted into executable code for projects, also so I can get working demos faster...

but now I'm running into the issue where art assets is the main limitation. I want lots of art for interactive scenes and unless its a project an artist is passionate about, I know it can be challenging to get a consistent source of art, hence the past 8 years of trying to learn art and trynig to figure out a work flow. While I have started to hone down what it is I want and can do with my art, its still the simple issue of not being able to spend enough time on a project each month. Which is why I can see how it is important to be able to work on a project full time rather than with just free time... which I'm conflicted about since I finally started my career as an engineer and really enjoy the job and I kind of need a stable income for the next few years since by the end of the next summer I'll have to make some major changes to life (need to move and buy my first place).

The one thing that really help is having enough money to commission writing to help progress the project's story while I focus on technical stuff.

Ooof. Hang onto that job, man.

 

[ririmudev]

What are some of your bottlenecks, and how did you widen them

Nice

 

[MidnightArrow]

I'd love to hear more about your pipeline, in general.

You must be registered to see the links

 

[CurvyLinesEverywhere]

You must be registered to see the links

Dat staircase though!

That's actually a really great tutorial, thank you. Very detailed. I like how you drill down into the differences between the programs, and leave some breadcrumbs referencing other workflows. If I'm not worried about exporting back to Daz, I still learned some things that will generalize to a MHX or Rigify workflow. Great stuff!

And as a nice side-effect, some of the people making money doing this are apparently now visiting your Patreon.

You must be registered to see the links

I'm actually kind of astonished at how much better your Blender render looks than your Daz render. Was this Cycles or Eevee? I'm so accustomed to being floored by Daz's HRDI and then needing to make up the difference in Eevee, but maybe the Daz render is just relatively flat because it's an indoor scene with no Environment lighting?

 

[CurvyLinesEverywhere]

Rendering anything is time consuming, Daz is render heavy and requires extensive prep, from models, areas, props etc. There's really not much of a comparison in terms of time saved between HS and Daz, as realism is where comparison ends. You really don't need to do any rendering with HS or KP.

It's basically taking a screenshot of what the realtime game engine powering HS or KP was already doing. So, like a low-rent version of Eevee. Fewer features, and most of the content is fan-made or fan-converted, but all you have to do is mess with the sliders and click Save Picture?

 

[CurvyLinesEverywhere]

Some people are snobs for visuals. They believe that 60 FPS is the "standard" ; required, but the reality is HQ or high res "look" from FPS doesn't come from FPS itself - it comes from "detail". In other words how detailed you can make something look in combination with a higher frame rate.

It does look better and more fluid up to a certain point, but the human eye can't really register anything beyond 40 fps in many cases. 30 is the minimum where people start to complain, 40 is the point where it really doesn't matter too much. 60 is "seamless" for the human eye, and anything else is really just an illusion. This is why people are leaning towards being able to display more intricate details in the quest for "high quality".

I bring this up because, it could be possible to also render "high res" images, boost the FPS to 40+, and still compress the files to get the same sort of next gen look. Obviously you'd want to lower your compression to maybe 70% (30% compression, retaining 70% of the image), but you could still get away with it and no one would really know the difference. This would probably be the optimal way to get a high quality look for your work without really needing to go the full mile. And most people wouldn't notice or rather - couldn't notice.

Sorry. When I said "the realtime game engine powering HS or KP," I didn't mean the output would be a 60FPS video, I meant the realtime 3D renderer that powers Honey Select itself.

You must be registered to see the links

That's a real-time

You must be registered to see the links

. It draws 3D models on the screen in real-time. It just isn't being used for what we would call traditional gameplay.

In other engines, there are TONS of things other than output image size that affect how long it takes to render an image. In fact, sometimes rendering an image extra-large and then scaling the image down to your target resolution can actually be a shortcut when you're working with Cycles or Iray, because it lets you get away with fewer iterations:

You must be registered to see the links

Eevee is a real-time renderer in Blender. It's probably generally a little slower than Honey Select because of Mesh Subdivision and Post-Processing and Color-Correction and other features with "overhead," but it's still leaps and bounds faster than a ray-traced engine like Cycles or Iray. I like Eevee because I think it offers a good compromise between speed and control, but your project's requirements might be different.

 

[MidnightArrow]

I'm actually kind of astonished at how much better your Blender render looks than your Daz render. Was this Cycles or Eevee? I'm so accustomed to being floored by Daz's HRDI and then needing to make up the difference in Eevee, but maybe the Daz render is just relatively flat because it's an indoor scene with no Environment lighting?

The reason is because Iray uses the sRGB color space (which sucks) while Cycles uses the filmic color space that doesn't blow out bright lights.

 

[CurvyLinesEverywhere]

The reason is because Iray uses the sRGB color space (which sucks) while Cycles uses the filmic color space that doesn't blow out bright lights.

Mmm. I can never get good results with Filmic in Eevee. Everything just looks vaguely muted and grey.

 

[MidnightArrow]

Mmm. I can never get good results with Filmic in Eevee. Everything just looks vaguely muted and grey.

That's because you're supposed to tonemap it with the compositor or an image editor.