VN Ren'Py The Moorcock Incident [Dev Thread] - Big Titty Snobbish Girl Slow Corruption

[osanaiko]

NMMMN Thank you for the suggestions, I appreciate your input.

You could do the 2D character rig method though less ideal for unique poses, it cuts down animation time by having them parts using, making verts and skinning them to bones which allows for IK (inverse kinematics solvers for the bones), or simple rotation, squash and stretch per part

I've already tried the rig method of animation for use in the "minigame" portion of the project, using a couple of different tools: Synfig Studio, Cartoon Animator, Inochi Studio. I'll preface this by admitting that I'm not really experienced in any type of art creation, let alone 2d animation. So a lot of my difficulties probably come from my own lack of knowledge and skills. However, here's my assessment of those options I tried:

- Cartoon animator is a whole big complex package of software, aimed more towards making actual TV quality (cheap Chinese kid's TV animation) stuff. I found it quite hard to get over the learning curve, and gave up quite quickly.

- Inochi Studio initially seemed cool, but once trying to actually achieve even basic stuff, it was revealed to be kind of incomplete as a general 2d rigged animator. The only part that seems to be getting development love is for doing those "hololive" style motion-capture-synced anime heads. I didn't get far with this tool.

- Synfig Studio is the tool with which I made the most progress. I posted an example of a handjob animation created with Synfig earlier in the thread. Unfortunately, the program itself it one of those ancient, creaky, open-source projects that only has one developer, and there are so many "sharp edges" on which the user cuts themselves over and over. Reminded me a lot of GIMP. It also is quite crash prone, and I lost in-progress work a few times. Altogether I'm not sure I'd have the patience to keep with it beyond what I made so far.

Apart from those tools there are of course some others... you've suggested Unity - which is a whole different box of fish.
I have spent a bit of time historically with Unity, and even today still use it when collaborating with the editing work for "A Long Journey", so I have at least some familiarity with it. But I'm not going to use unity for 2d game dev, it's like using a demolition hammer to tap a nail into a wall. Or maybe using a 3d-printer to make all the parts for a lego set.

Other candidates would be the "OG", Live2d, or maybe even Blender given the recent work to support a 2d workflow. It would also be theoretically possile to do something with pure Renpy ATL, but setting up e.g. pivot points etc. by hand would not be fun.

All up, I don't think that rigged animations are what I want for the story scenes of the game.

I might keep working towards using the technique somehow in the Minigame sections, rather than frame based animation.

 

[osanaiko]

I'm sure you are all dying to know what I've been up to recently. I won't be shy, I'm happy to tell you: I've been experimenting a bit with Video Generation models, specifically "WAN2.1 I2V".

So far the stuff I'm made is kinda cursed, and there really isn't much direct control about what goes one in the shot. But if things progress as they did with the static image Stable Diffusion tools, then hopefully we'll get some seriously effective tools in the near future.

I've spent 4 days on it so far, and will probably put it away for a while as I don't think the technology is ready yet.

For anyone interested:
- I followed a guide I found on Civitai to install ComfyUI portable, and then used an installer script to setup all the required custom nodes, and finally a Comfy workflow. All were created by "UmeAIrt":

You must be registered to see the links

- then downloaded a bunch of models (WAN model 18GB + 18gb + 14gb, CLIP model 6.5gb, Clip vision model 1GB, some other stuff).
- one downside is the huge VRAM requirements when using the least-quantized model (wan2.1-i2v-14b-480p-Q8_0.gguf). Even with 24GB VRAM, I can only run this tool if everything other than one copy of firefox with one tab open (for the ComfyUI) is closed. Note that there are partially quantized model versions (i.e. wan2.1-i2v-14b-480p-Q6_K.gguf and others ) which use less VRAM memory. There are also some "vram-poor" versions on WAN2.1 around that fit even on 8GB cards, but the quality is much much worse from what I've seen)

There are LORAs for guiding the video output, but so far they mostly seem to be straight up hardcode porn, or visual style for specific game characters. I really wanted some that specified certain basic actions like "walking away from camera", but there's nothing like that which I could find in Civitai.

Here's an example or two of what I've managed to make so far. It's still early days, and I'm not yet certain that this is going to be useful to me - it's very time consuming to get a good output, and even then they are short and jerky, :

1. In order to avoid dealing with the face features mutating (as I don't have a LORA for the visual style), I thought' I'd use a rear view. But then the model decided it knew better and flipped the character around, revealing a barbie doll body.... Also note this one is 12 fps and at low resolution. The workflow does allow both upscale and interpolation, but I hadn't enabled it here.
View attachment 225850_OG_00001 - Copy.mp4


















2. Here's one with the camera in front, using a non-canon test image as the source (short hair Juliette is not good). However I just could not get the model to make her smile!
This one I has both the upscale and the interpolation enabled, so it's 32fps and 640x960px. Muuuch smoother.
View attachment 185108_UP_00001 - Copy.mp4

 

[osanaiko]

And finally... (three posts in one day, what is this madness!)

I did an assessment of where I'm up to, and what is left to do. I'll post the details tomorrow.

 

[osanaiko]

As I mentioned the previous post, this past weekend I sat down and did a full assessment of where I'm up to. The dev process has been quite scattershot so far. All of the initial burst of "new project yay!" energy went into making the minigame proof of concept, and since then it's been a lot slower as I work on stuff (scene art) that's neither in my existing skill-set nor especially enjoyable for me.

So far this is what I've got built in to the game itself. These make up most of the "chapter 1" of the story:

• Scene 0: Game opening intro - DONE
• Scene 1: "Meet FMC" - DONE, but I want to got back and polish a bit.
• Scene 2 "Exposition of FMC's situation" - DONE
• Scene 3 "Huh, that's weird" - DONE
• Scene 4 "Rescue action montage!" - not started, been experimenting with chibi art
• Scene 5 "A wild Donut appears" - DONE
• Scene 6 "Antagonist character intro and some exposition" - plot outline only. Scene Background done, will mostly be able to re-use sprites.
• Scene 7 "Meet the antagonist's side kicks" - plot outline only. Will reuse the corridor background asset. Needs more character sprite art of the 2 sub-antagonists.
• Scene 8 "Somethings wrong, i can feel it" - plot outline only. needs new art for everything.
• Scene 9 "Oh shit! It's crisis time" - plot outline only. Not sure what is needed, I haven't planned it out yet.

And then there's all the other assets and stuff which I've made, most of which will get used in the game eventually:

• gloryhole minigame proof of concept, in a separate renpy project
• A partial implementation of a "quest" handler: i.e. "go to location / click on person or item to trigger next scene" events system
• map and "location" movement code - still a bit immature, will need more work to integrate with the quest manager
• a couple more of scene/location backgrounds.
• lots of concept / "maybe useful" generated images. I've found myself revisiting and pulling stuff out of these archives quite a bit, so it was a good decision to not throw away all the "close but not quite right" images.
• some AI gen music (one track used in the into, and another which I quite like that should work well as one of the minigame background musics)
• some AI gen animations (tests / proof of concept only)

So, that's a bunch of stuff that is partly built, but not ready to be used without more work.

Next: What is the path forward? Well, I think I probably need to at least finish the following to get to the first release:

1. Complete the 5 not-yet-done scenes from the list above
2. Add a couple more scenes after the crisis to connect to the first minigame part. Got some ideas for these.
3. More work on the Glory Hole minigame to make it an actual "game", similar to the inspiration in OnEdge. Right now it's just a dev harness driving the action animations etc. Will need more work scripting and balancing the available "moves" and the costs. I'll be interested to hear feedback from alpha testers to understand if the UI animations etc are working well enough. If there's too much stuttering or lag I'll need to look into using Renpy "Creator Defined Displayables" (CDDs).
4. Currently the scenes in this Chapter 1 are all kinetic, with no meaningful choices for the player. I will add at least some choices so there's a feeling of self determination for the player. But I also don't want to go too far here, because choices -> story branches -> even more time to make progress.
5. Finally, I'll run some closed alpha play tests to get feedback.

I very much understand that keeping stuff to myself means I am not getting any feedback - I've certainly preached to other developer's that this is a terrible idea. But I'm very hesitant to make a "release" without at least some sex-scene payoff - I've read plenty of grumpy feedback from forum posters/reviewers that a game without sex is pointless - and I agree, why else are we all playing this stuff. Unfortunately, due to the story design, it takes a while to get Juliette on her knees...

How long is it all going to take? I honestly can't promise anything. It's certainly been far longer than I expected originally, but I guess I was just naive. My living situation (and full time job) means I get at best a few hours a night of free time to play adult games or work on not-safe-for-work art for my own adult project (text is much easier to work on without risking problems, but text is not the thing holding me back for this game...). And some days after coming home from an "exciting" time at $DAYJOB it's very tough to sit down and grind on artwork with those few hours of freedom.

The only thing I can commit to is that I'm not going to quit. The good ship Moorcock shall sail towards her destiny!