Extras | [gif version] [2] • Source: Yonia
Typically the bigger they start with, the bigger you can go without it getting too distorted. That means the end result here could go pretty big.
Typically the bigger they start with, the bigger you can go without it getting too distorted. That means the end result here could go pretty big.
Not really a fan of how it turned out, but using the puppet tool is faster than using paths. I don't have experience picking the right spots with the tool.
One thing I did that doesn't really show up, but always bothered me when doing drawn things is the line art. I removed the line art around the chest, then put it back in after the animation that way the line art doesn't increase as her chest does.
I had the file labelled as Kipos, but I don't know the why.
This took forever to rotoscope and I really got bored doing it. But I was really happy with the expansion code and things, it really made that part so much easier and the result was really quite good. I'll make a post about the expressions later.
I had the file labelled as Kipos, but I don't know the why.
This took forever to rotoscope and I really got bored doing it. But I was really happy with the expansion code and things, it really made that part so much easier and the result was really quite good. I'll make a post about the expressions later.
I've been working on a different video and it's been difficult to work on - took a quick break to make this one. The reason the other video is hard is because the lady turns from side to side. If you look at most of the videos I've done in the past I don't do ones where they rotate much. When I create the expansion shape it would have to be recreated on each turn and keeping them the same size/shape is really, really annoying.
So I set out on writing a new expression to auto-scale the rotoscoped shape, and this led me down a path of how to then weight certain points from being scaled as you don't want the top of the chest to scale upward, or really at all. Anyway, I've coded that and it's working... now I just have to continue to rotoscope the footage. I think I'm 25% of the way done with that, so it's going to take a bit.
Next one is a video though. I have another video I'm trying to figure out how to do, that's what this one was supposed to test for, but it got too messy.
I did use stable diffusion to generate a single frame, then it transitions between the original size breasts and the new much larger ones. Then a lot of animation work goes in to animate it correctly so it looks like they're not just an image pasted on.
I always wanted to do a growth in a shirt that streches it to the point of sliding out.
I'm sorry there's no jiggle, but I really just focused on the growing out of the shirts. Redgifs compressed it pretty hard, but the left shirt gets thinner as it streches out, was pretty proud of that. Perhaps next time I'll figure out how to do it while a bra is being worn, that'd be cool.
There's a faster version in the mega folder if this one is too slow.
The first 1 second of the clip is largely unaltered, but then she slowly grows into a larger pair that is AI generated. It's actually only a single frame of the breasts animated to look like they're moving.
The image was made awhile ago, but I remember this being me messing with control net and using a base image. And after making the other two, I learned what to look out for with inpainting and how it comes up when animating in after effects later.
As a last note... you know how many iterations I had to run through to get it to generate good hands?! That's a real thing in stable diffusion, hands...
This originally had like a bra/lingere top, but I didn't do a great job with the inpainting in stable diffusion so I gave up and just went without.
Tried for a heavier jiggle, but kept it simple without stacking a ton of effects.
I hope to make more things this year, I mean, they take a long time, but my goal is at least 1 thing every 3 weeks. Sometimes things take 3 or 4 days, other times doing a whole video can take upwards of 2 weeks.
Last year I started messing with stable diffusion (AI image generation) and that took some time to learn and get used to. I think it works well with what I do and I feel a lot more comfortable creating the images there than just finding them online, especially for attribute theft stuff because someone was always the loser and that just made me feel bad - but now they're not real people. It can't do video just yet, so full video will likely just be supplemented with AI.
I'm not going to work on any After Effects script anymore - sorry. The script stuff just became too much work and I lost interest in doing it. I think I've got all the images in the mega folder, if a recent one is missing let me know and I'll track it down, old ones are probably lost to graveyard of the internet.
I have 2 things I'm going to post shortly, they were both test cases for stable diffusion and the new version of after effects. Getting back into the swing of things. Then 2 more have been scheduled for this week. I'll decide then if I should post things to reddit... test the waters so to speak. Why not just post them all? Gives me more time to start working on something else, spreading out the content.
I finished this awhile ago, but I didn't post it because I wanted to do some effects with the phone. Then I just lost motivation to do it so here's where it was left.
This has a different effect than I normally use for the bulges, normally I use the Bulge effect, this one is the glass effect with a lot of the options disable (otherwise it tints the colors).
There's a couple other moments in the video that are also good to edit, so I may get to those eventually.
I tried a new masking approach and it worked fine, but made the morphing unstable. I spent more time trying to salvage it than anything else. Since I haven't posted anything in awhile and it took so long I didn't want it to go to waste, but it's not super great with the jumpy movement.
As a side note, I tried Davinci Resolve (fusion), and while it seemed like it could do a lot, I'm not sure this type of animation is easy to do in it.
The more disappointed she is, the bigger they get... so pretty big.
I wanted to do this one because the blur at the bottom looked fun to animate, then it ended up being not really noticeable at all, oh well. I animated the chest twice, this was the larger of the two and I'm glade I went with it. The hand does animate a bit, but I got tired of painting in the background so I didn't do much of anything else. The bikini was a tricky setup because I wanted the strings to move correctly, but turned out you just rely on puppet pins and it just works.
As with all the recent things, this took more time than I wanted but I finally finished it. From the arms that cross over to the very busy background, there was just a ton of cutouts to mask out. The shadows/highlights are pretty good here, still not perfect but overall I'm happy with them. So the entire torsos are actually recreated, so the skin texture, lighting, shadows, tone, etc all had to be recreated for both directions. Originally the left girl is pretty flat, and the girl on the right is not flat. I wanted to go much bigger, but I ran into an issue where the left lady twists her body too much, so going bigger made it much more apparent it was fake.
Update: I went as far back as 2015 fixing the posts, but redgifs really screwed up an update. iphone users seems to be hit or miss as to whether or not things load. Not much I can do, sorry. Image/Video hosting isn't exactly something there's lots of options for.
The problem with using liquify near the edges is that you end up creating gaps on the edges. So you either can't use it near the edges or you have to create the background. I thought it was clever, but I used an echo effect and put it behind the original frame. Basically let me work with both the last known good frame and upcoming recent frames near the edges so when I used the liquify effects it didn't create holes.
• Source: ???
Link to folder: mega.nz folder I'll need to make a page here explaining the expression I used, but basically you give it three keyframes: start, sling, and end. The middle keyframe (sling) controls how fast and how far the sling/bounce happens. It works pretty well consider it has so few options. If you open the .aep file, the expression is in the Assemble comp, on the LEFT CHEST layer, under the "horiz" and "vert" effects. I also made a little test zone in a comp for people to just play with.