VN Ren'Py AutoMating [Development Thread]

[dabberman14]

Hello guys, I am making a NieR Automata Ren'Py game and I want some feedback.

Spoiler: Who am I

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: My plans for the game

You don't have permission to view the spoiler content. Log in or register now.

If you disagree with any of these ideas or have some of your own, feel free to tell me all about it. I do not have an artist's soul, and I won't be discouraged. But I am stubborn, so you might just remain unheard.

Spoiler: My Questions

You don't have permission to view the spoiler content. Log in or register now.

I'd also like some feedback on what I already have in my renders, such as if there are any glaring issues with the geometry and textures as well as whether you feel I'm using too much post processing with the glare, bloom and dust. Tell me if the shaders look right, I know there are some issues, I wonder if you notice them.
Also on the prototype UI I have lightly modified from the default in Ren'Py. I also have a few questions regarding usability and my deranged code. I do have a working prototype build I would like to share for feedback, but I'm not certain of the rules of the forum for uploading random untested builds.
Keep in mind everything is subject to change.

Also please give me your ideas, any and all. Without a creative bone in my body I need all the suggestions I can get.

 

[Saki_Sliz]

Ah, its so nice to see a well formatted post.

I just want to make a game of a good quality.

Same. started making art 8 years ago so I could start making my own games... still no game but I've learned a lot and I'm still on my way to... something. Anyways, as you can imagine I have a lot to say. But mainly what I want to say is encouragement and some suggestions to consider to hopefully keep you going.

Spoiler: Answering your questions

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: A thesis you didn't ask for :)

You don't have permission to view the spoiler content. Log in or register now.

 

[dabberman14]

Impressive, you have me beat on the wall of text, very nice. However...

Spoiler: Some replies

You don't have permission to view the spoiler content. Log in or register now.

Thanks for answering. Reading all that has taken my precious time. Now I will go to sleep. And if there is an again, I will conserve my time rather than enagaging in the word wall contest.

 

[Tripzy]

Why not just post video links? I find it a bit sussy to open a random webp because of the way it can be used as an attack vector.

 

[dabberman14]

Why not just post video links? I find it a bit sussy to open a random webp because of the way it can be used as an attack vector.

Lmao. In that case you really should not be using Google. Or download any Ren'Py game on here. Or use a web browser at all. webp is the modern replacement for jpeg and png, pioneered by Google and adopted by all large scale web-hosters due to its bandwidth efficiency.
But you don't have to believe me. Just for you, I converted them into jpeg. They aren't videos, you're thinking of webm. As you can see the file size has increased, despite the image quality being lower.
Please refrain from posting disinformation in the future, thank you.

 

[Saki_Sliz]

I am not aware what going beyond Ren'Py means,

Mainly, I talk with some devs and they want to add a game mechanic that's not so easy to add to renpy. But this would be something like making the game into a side scroller. People can and have done it, using sophisticated transform tricks. but I find that its much easier to use something like unity/godot because you can immeditly start coding the idea from scratch, where as with some renpy systems you have to work around what renpy can and can't do (such as reusing the transform system to fake a side scroller game)

Cleaning up what I got would take 90% as long as doing it from scratch so I abandoned the idea. Was my experience an outlier?

Nah, lots of assets don't port over well through FBX, a lot of the idea behind a 'workflow' is just gaining experience with what does and doesn't work, in some cases that would be trying many assets and seeing which one's are usable vs not usable.


All in all it sounds like you aren't too unprepared, expecting something rough at first but looking for ways to improve as you go without spending too much in one area. Best of luck! I look forward to future updates!

 

[dabberman14]

expecting something rough at first but looking for ways to improve

Most likely what you'll get is some nice renders and not much else. Ways of iterating on 3D seems more or less obvious to me, meanwhile figuring if my plot or characters are any good seems problematic, yet possible. How to fix the two however... Seems completely ephemeral, inscrutable. I might be able to figure my character is bland and lifeless. But I have no idea how to fix this while keeping in line with the rest of the writing. I will put something together, maybe it will even be good, but I can not foresee. Writing a short story with some minor playout followed by a non-conclusion is easy and I find it quite fun, expanding on it seems an impossible task. We'll see though, perhaps it will all just come together on its own (yeah).

 

[dabberman14]

Spent some time learning Marvelous Designer for the cloth physics, but couldn't figure out how to recreate 2B's dress faithfully, it does defy gravity after all. So instead I reused the game mesh and made my own textures in substance painter, took way too long. Tell me what you think. I wonder if I should just use the velvet shader instead.
Turns out the website may re-encode .jpgs, but seemingly doesn't for the .webp, at least in this case.

 

[Saki_Sliz]

Clothes generation using MD or blender is its own secret cult. I can only do simple things, the physics is great for small details, but for actual designing clothes, I don't know how artist manage to get little nuances, such as the shoulder's on 2B's dress.

I would say the results look stellar. I would be worried about reusing the original mesh, but since you mentioned this was a renpy project, I imagine you'll be doing renders, where images are considered a transformative work thus any copyright concerns regarding the source mesh wouldn't be an issue.

rigging the dress might be tricky. I know daz has a transfer utility, but it always destroys the mesh in some way, and it really sucks at doing dresses because of this (DAZ tries to calculate the rig by morphing the dress to match it's base figure, and then tries to re match it to the target character). you may have an easier time if this dress was one from a smutbase figure, already rigged, but if not from smutbase, I imagine that there may be some models of 2B for both smutbase and daz that you could get, but in both cases you would probably need to move everything to maya or blender just so you can make any scultping adjustments to get the dress and or 2B to fit each other. If the dress isn't rigged, I would suggest looking at blender's weight transfer function, that allows it to try and match the dress to 2B's rig, without adjustment (unlike daz). If you do end up using blender, such as for all the rigging issues with dresses, then you probably don't need to hasle with trying to port the rigged dress to other programs, you could just continue to learn and use blender for the rendering.

As for the velvet shader, don't worry about it until you've got things together and are starting to make test renders in the final engine (blender or daz or other). I tend to avoid the velvet shader all together. Unless you are making a very specific scene and know what you are doing with lighting (only really works for night time scenes), both daz and blender's velvate shaders suck. It's not that they are bad, but the velvet lighting effect is special, and only really makes sense to our brains if we see the object moving or are familiar with the object being velvet (such as a valley's rope), because velvet in general is a strange lighting effect that doesn't make sense to our brains unless we have enough data to understand what we are seeing. using velvet shaders just make it looks like the object has a bad lighting/shader effect. I find that both daz's and blender's ability to add a bit of a sheen effect (light wrapping around) is usually good enough to show off a 'velvet like feel' to the material, usually the white outline you see on the right side of your dress image (front view)

yay progress!

 

[dabberman14]

I don't know how artist manage to get little nuances, such as the shoulders

The upper part is doable IRL, not sure about MD, but with some finnessing should work I think. The lower part cheats a bit, but still possible. The problem is the skirt. It's not possible to have that big of a slit and yet still have such large wavy folds right on the edge here. There is nothing for the fabric to lean against except the seam all the way above. Strengthening also makes the waves not as pronounced, I'm sure someone closely familiar with MD could pull it off, but this dress is physically impossible without a skeleton.

If the dress isn't rigged, I would suggest looking at blender's weight transfer function

Yeah, I'm using Blender, I'm hoping the weight transfer will 'just work'. This is rendered in Cycles, which will be a problem for animations but I might figure it out. I did my best de-triangulating the mesh, which isn't too great but should probably be good enough, never done a weight transfer, but can't be that hard, right?

As for the velvet shader, don't worry about it until you've got things together and are starting to make test renders in the final engine

But the outdoor scenes above are also in Cycles! But yeah, Velvet BSDF looks wierd, needs work, but so did this. Each material is at least 50 nodes, many are likely redundant, but it's the best I could do. The white outline is a combination of a very high sheen value with additional fresnel effect mixed onto the base color with an add node. The dress is also ~50% metallic as that is the only way I could get the shading right, and it is also how the game model imports into Principled BSDF.

 

[Saki_Sliz]

The dress is also ~50% metallic as that is the only way I could get the shading right

Yep, had to do that a lot, makes things less bland. but all and all I think it looks really good, I probably wouldn't notice any issue material/texture wise, its going to be cloth clipping that's the problem.

Daz has a nice smoother that automatically brings clothes to the skin if the clothing clips a bit, a similar thing can be done in blender using shrinkwrap, and setting it to 'above' or 'outside' mode so it only shrink wraps parts of the mesh that clip into the character, you can then layer on corrective smoothing so that the shrink wrapping doesn't look noticable, and this helps the clothing retain details such as folds. just little things that can make the posing and animating process easier.

 

[dabberman14]

While making the dress I kept looking at the gamerip model for reference and eventually I just found the rigged model from smutbase (which is based on DAZ G8F) I downloaded a bit exaggerated in its proportions. Mostly the pencil thin waist and legs. Not to argue in favor of realistic proportions or modern sensibilities or anything, but I do find it kind of laughable. I understand those kind of forms is probably what a regular porn watcher enjoys, yet still I found wanting to bring the body more in line with the gamerip model. On the other hand, while not visible under the hair, the smutbase model's head and face is too puffy, I slimmed and remodeled the brow, eye, jaw and cheekbones to be closer to the original japanese design.
I added fiber brows on top of the textured ones and replaced the alpha plane eyelashes with curve fibers as well. Most of my time as been spent converting the hair mesh into hair curves, which proved to be an extremely laborious process. Each and every strand of the original mesh has been converted to hair particles, if increase the density, it will look almost the same as the original mesh hair, except also crash my PC. I am somewhat of a hair fetishist so making it quality had to be a priority, despite the many downsides this brings. Yet, unfortunately the result doesn't necessarily seem qualitatively better than the original to me. I am somewhat stumped by what to do here as I do not really know what else I could try instead to make hair better, so tell me which one you prefer here.

Spoiler: Comparison

You don't have permission to view the spoiler content. Log in or register now.

 

[Saki_Sliz]

Yet, unfortunately the result doesn't necessarily seem qualitatively better than the original to me.

The. Worst!

ok after reading, sorry about this, but I have to give you the 'it depends' answer, and give you the engineering/technical artist response.

Spoiler: Clothing has an impact on body perception

You don't have permission to view the spoiler content. Log in or register now.

Second, judging which body is very very hard. I don't just mean its an issue of subjectivity. I'll make a character Friday, experiment with proportions. I'll save everything, use them as a base sculpt for another figure. But then I'll come back after working on a different style of character, I'll look back at my old design and go 'what was I thinking with those proportions.' and overall be confused how I thought It was ok or good. So between the two models I won't discuss which style is better.

However, I won't leave you hanging. I can give you a way to measure which is better!

Spoiler: Body type dependent on the type of scene shots you are doing

You don't have permission to view the spoiler content. Log in or register now.

If you are going vn, for nearer shots, you may need to explore more lighting techniques and ideas.

Spoiler: Lighting and skin

You don't have permission to view the spoiler content. Log in or register now.

So I would say, use your model if you like it, keep the other in the back of your mind, but always be ready for if you want to experiment and change things after learning a bit more.

Spoiler: Hair

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Eyes

You don't have permission to view the spoiler content. Log in or register now.

 

[dabberman14]

Spoiler: Clothing

You don't have permission to view the spoiler content. Log in or register now.

between the two models I won't discuss which style is better.

Oh right, sorry I wasn't very clear. I am actually very happy with how the body turned out. It is the hair that I center my doubts on.

Spoiler: Shot preference

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Lighting

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Shaders and drivers

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Auto specular

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Hair

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Eyes

You don't have permission to view the spoiler content. Log in or register now.

Thanks for the feedback! I really appreciate being able to talk to a knowledgeable person! Much of this would be so much easier if I had a better idea what it is I want to make.

 

[dabberman14]

Found a way to make hair render quite a bit faster so I increased the density. Looks better I think, but the sparse look has its own charm too. Also made the sun softer.
View attachment outputVII.webm

 

[Saki_Sliz]

Much of this would be so much easier if I had a better idea what it is I want to make.

The struggle of everyone who ever wanted to make anything XD

Spoiler: Composition

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Outdoor lighting

You don't have permission to view the spoiler content. Log in or register now.

see you are the one who made that complex skin shader on Smutbase I came across a few months ago.

Spoiler: Gasp! Yay, yah found it!

You don't have permission to view the spoiler content. Log in or register now.

About specularity, I've searched around, but it doesn't seem like anyone really grasps what the slider actually does beyond "might make things more or less shiny".

Spoiler: ta da~

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Hair

You don't have permission to view the spoiler content. Log in or register now.

Thanks for the feedback! I really appreciate being able to talk to a knowledgeable person!

Glad I come off as knowledgeable
just doing this as a hobby, not a profession, so I'm a bit all over the place
anyways, sorry for the delayed response. I figured I would spend time typing a reply the moment I started reading, so I waited till I knew I would have the time