Did a smaller render to show a different perspective. No time today to actually work on stuff, though.
Daz to Maya Workflow (and apparently Dev Log now)
- Thread starter nillamello
- Start date
I followed your path (and quit multiple times). Great job man. I have to say though.. unless you get something like redshift for maya i cant see it being worth it. And at our level (people who dont have 50k in a render farm account) cinema4d is better. I want to make a movie honestly.. but im not willing to drop to mayas quality level. I can build and use everything in daz 1000 times easier. the cost is.. not having some tools and using Iray.. which is 1000x better than any cpu renderer. Its hard to go from irays stunning photorealism even at basic levels to the standards cpu renderers are willing to accept. I would love it if the industry would actually develop standards and co-support a pipeline tool making all formats truely interchangeable ( of course you'll lose things engine to engine).
Arnold has a GPU version in beta (available with the newest release version), but it crashes on my machine... I think my semi-old card just isn't quite compatible and I haven't bothered to check the newest version changes to see if it's working properly. Lots of people are happy with it, though, apparently, it's quite fast.
I think Daz is the place to be if you want photorealism (though I'm not personally familiar with C4D, I've heard good things). I think of Daz as a purpose-built solution for photoreal renders of people, and it does it really well. Because of the focus, though there are a lot of tradeoffs to be had in other areas of the renders (I think viewport efficiency and GRAM efficiency are probably the largest hits, but also all of the small things that can detract from realism, like body physics are lacking), which limits environments detail and number of characters, and so forth.
Let's see, for a realistic short film... Grab your characters from Daz, animate them in your chosen program (Maya for me, but there are a ton of good options), grab walk animations and basic gestures from an
You must be registered to see the links
or from the Unreal store or something, clothing is an issue for good simulation, so you may want to try Marvelous Designer first for basic movements and then use whatever program you intend to build your physics simulations in (Houdini?) to do the final cloth physics. From there, import all of the cached files (I'm assuming it will be alembic geometry caching, but your chosen render program may have a better option... I'm still talking in Maya) into your final program of choice. From there, you'll need to tweak textures and stuff, make sure the shaders are right and all that. Something I've learned in this process is that I REALLY want to get everything looking right before I animate, and that is REALLY WRONG. The scene should be as empty as possible. Characters don't need hair, clothes, facial expression, textures, or ANYTHING. Just animate them clean, get it right, and THEN make them pretty. Anyway, from there, you can add all of the simulation data, the clothing rigs, hair (I recommend finding a hair solution that you really like... I've grown accustomed to Xgen in Maya, but there are some other respected solutions like Yeti or Ornatrix), and all of that. A single program that excels at every aspect of production is still a pipedream... some of them have plugins that greatly help (the new Bifrost graph looks like it may finally get Maya into a good place with particle simulation for example), but still... they're all built to import and export into other programs (Daz being a bit of an exceptions, since they refuse to export map data) so it's not a huge inconvenience. I find that it can actually be sort of helpful to have a limitation on what you can work on at a given time. If I'm in Marvelous doing a cloth sim, I'm going to stay there until it's finished. I'm not going to get distracted with an animation tweak or anything like that.Anyway, that's my two-cents. I went into this never intending to do photorealism (it's one of the reasons I left Daz, since I just couldn't quite get the look I wanted), so I'm not entirely certain how to get that result elsewhere. I wish you the best of luck and I'll be interested to see what you come up with.
- Aug 15, 2019
- 153
- 143
I heard that Daz Studio rigs significantly rely on daz morphing system. So, animations in daz look better because they use morphs to minimize self-intersections.
Yeah, but the cool thing about the latest import utilities is that they actually also import the joint-reliant morphs (the flexing and so forth that happen when you move various joints), so a lot of the natural look is retained.
I've mentioned this in posts elsewhere, but I'm not sure if I've actually said it in this thread much, but if you want to make truly realistic characters in Maya, you need to put in a ton of additional work beyond importing. Daz is a purpose-built application to do one thing really well, which is real people. You can certainly use the system for stylized purposes, but the core of the product is realism, and that is also part of their proprietary intellectual property. Sure, they let you export a good deal of the model into other programs, but the really important bits stay in Daz and only Daz.
Maya (and other programs) have a lot of flexibility, but also have a learning curve. For example, if you want fully dynamic self collisions, muscle flex, skin stretching, deformations and so on, you can use Maya Muscle to create an actual flesh and blood human being from the ground up with realistically placed muscles that flex and bend according to the animation or environment. Or if you want to go all-in, check out Ziva Dynamics for some truly crazy next-level shit. But that again is the selling point of something like Daz: half of the options with none of the hassle. I had an easy choice since my project is not photoreal (I'm aiming for the Frozen/Tangled type of look). The vast majority of the time, I don't really need to worry about realistic muscle movement ("looks real" is good enough), and if I need something to be squishy, I just make it a softbody and move on.
Honestly, there is a proper program for every project, and a lot of people are going to find that Daz is that program for them.
Been working on environments lately, but now that the kids are back in school, I can return to the sexy bits without having to tiptoe around hoping they don't burst into the room to s computer screen full of digital cocks.
MC's room (this is version one of a few different layouts based on future choices).
It's pretty bare and dark to start with, but the player will be able to add additional furnishings and so forth. The mirror acts as a sort of player settings screen where you can see your stats (which won't actually be implemented in the first build, since the stats won't matter...
).
Mirror close-up:
Yet again, basic and dark, but he has a great view from his window of the crystal palace.
And then I have the home library.
I like this one. I was a bit worried about making the rug, but it actually turned out quite nice and was super simple. Just made a flat cylinder, put hair on it, fiddled with the thickness and length, added a little noise, and plugged in the pattern for the tiles on the wall as the color and boom, nice little rug.
And a night view of the same room with some glowy bits.
MC's room (this is version one of a few different layouts based on future choices).
It's pretty bare and dark to start with, but the player will be able to add additional furnishings and so forth. The mirror acts as a sort of player settings screen where you can see your stats (which won't actually be implemented in the first build, since the stats won't matter...
).Mirror close-up:
Yet again, basic and dark, but he has a great view from his window of the crystal palace.
And then I have the home library.
I like this one. I was a bit worried about making the rug, but it actually turned out quite nice and was super simple. Just made a flat cylinder, put hair on it, fiddled with the thickness and length, added a little noise, and plugged in the pattern for the tiles on the wall as the color and boom, nice little rug.
And a night view of the same room with some glowy bits.
- Jul 19, 2017
- 907
- 2,238
Not so much, no. The windows actually took the most time relative to the rest of the scene. That said, this scene did take a long time to render (three hours per frame, with four frames: morning, noon, evening, and night). My usual render settings rely on denoise layers to clear up render artifacts, but the process also took out a lot of detail (the rug especially lost the look of being actual fibers, but the biggest problem was that the windows just turned blurry). Rather than spend time in photoshop combining the denoised and raw images into something semi-nice, I just had the scene run again overnight with much higher cycle settings. I'll need to test this rug idea out in better-lit environments before I can really make a final judgement on total time impact, I think. I guess my initial take on it is that the actual render time is pretty quick, but the final product needs to be fully-rendered without post-processing to be usable. Putting the rug on a different render layer and combining them in post is probably the best option for animations.
That said, using the hair shader for this proved to be really versatile thanks to the color and specular options available. For example, making this into a silk rug would just require doubling the density, making the hairs a little thinner, and increasing the specular reflections for a matte-shine surface instead of the current rough yarn look. And since it's still just hair, if there was a tread pattern or footprint, you can just comb the hair down in places to create that look.
That said, using the hair shader for this proved to be really versatile thanks to the color and specular options available. For example, making this into a silk rug would just require doubling the density, making the hairs a little thinner, and increasing the specular reflections for a matte-shine surface instead of the current rough yarn look. And since it's still just hair, if there was a tread pattern or footprint, you can just comb the hair down in places to create that look.
- Jul 19, 2017
- 907
- 2,238
Yeah, the guy who made the plug-in went through all the effort to automate the morphs. Huge time saver.
Ooooooh kay.
I played a lot with looping animations, and then when I got it totally perfect, I ruined it and then rendered it!
So here's my process:
1: import characters and all that fun stuff, drape hair, whatever. Normal things.
2. Since it's a looped animation, I chose the start point/endpoint and went ahead and keyframed that pose in on both the first and last frames of the planned animation length.
3. Then I isolated all of the portions that came into contact (her head, his dick, his hand): and animated them together so that I could maintain consistency. So far, this really isn't taking long and I'm pleased with it.
4. While her head is still isolated (so that I can see all the angles without getting other things popping into the frame), I added pose layers for her facial expressions, pulled her lips to the correct places, added a blink, stuff like that. Initially, I had her tongue visible at the bottom, but then her jaw was so unnaturally opened that she looked like a monster.
5. Finally, I brought the scene together again and........ his leg is fucking RIGHT THERE. Like, his leg was so high that I couldn't even see her eyes, let alone the actual contact point. Fuck. So I pulled his leg down out of the way. And ruined my initial animation key. Replaced keys, corrected keys, twisted keys, fucking fuck fuck keys. That's why his leg is animated at all.
6. Still unknown to me, because it wasn't readily visible in the low-quality playback, one of his shoulders apparently was never keyframed into the last frame, so it was a floater. And thus, he does not smoothly transition at the loop point.
7. Hair. Initially, I used the preset I always use, and it looked nice. Draping elegantly over her shoulders and junk. Then I remembered that I'm in a loop animation, so I scrapped that look because it fell down at the beginning and couldn't be returned to its original location at the end of the loop. I basically made her hair fall naturally outside of the animation frames and starting at the animation portion, I increased the rigidity. There's still a tiny bit of movement, but I'm honestly not really happy with it. If I have the opportunity to tweak it later, I will, but ultimately it's a learning experience and the quality isn't horrible.
Also, somehow the initial downward thrust is much faster than it's supposed to be.... I must have done something wrong.
Edit: I also added a dynamic jiggle animation on her breasts. I forgot about that part since it was entirely invisible once I added the rest of the characters and the props into the scene. Go me.
I played a lot with looping animations, and then when I got it totally perfect, I ruined it and then rendered it!
So here's my process:
1: import characters and all that fun stuff, drape hair, whatever. Normal things.
2. Since it's a looped animation, I chose the start point/endpoint and went ahead and keyframed that pose in on both the first and last frames of the planned animation length.
3. Then I isolated all of the portions that came into contact (her head, his dick, his hand): and animated them together so that I could maintain consistency. So far, this really isn't taking long and I'm pleased with it.
4. While her head is still isolated (so that I can see all the angles without getting other things popping into the frame), I added pose layers for her facial expressions, pulled her lips to the correct places, added a blink, stuff like that. Initially, I had her tongue visible at the bottom, but then her jaw was so unnaturally opened that she looked like a monster.
5. Finally, I brought the scene together again and........ his leg is fucking RIGHT THERE. Like, his leg was so high that I couldn't even see her eyes, let alone the actual contact point. Fuck. So I pulled his leg down out of the way. And ruined my initial animation key. Replaced keys, corrected keys, twisted keys, fucking fuck fuck keys. That's why his leg is animated at all.
6. Still unknown to me, because it wasn't readily visible in the low-quality playback, one of his shoulders apparently was never keyframed into the last frame, so it was a floater. And thus, he does not smoothly transition at the loop point.
7. Hair. Initially, I used the preset I always use, and it looked nice. Draping elegantly over her shoulders and junk. Then I remembered that I'm in a loop animation, so I scrapped that look because it fell down at the beginning and couldn't be returned to its original location at the end of the loop. I basically made her hair fall naturally outside of the animation frames and starting at the animation portion, I increased the rigidity. There's still a tiny bit of movement, but I'm honestly not really happy with it. If I have the opportunity to tweak it later, I will, but ultimately it's a learning experience and the quality isn't horrible.
Also, somehow the initial downward thrust is much faster than it's supposed to be.... I must have done something wrong.
Edit: I also added a dynamic jiggle animation on her breasts. I forgot about that part since it was entirely invisible once I added the rest of the characters and the props into the scene. Go me.
Last edited:
Just throwing something up to say "hey, still here, still trying to improve". This is the first in a series of images that are larger than normal. I basically couldn't decide on framing, so I just gave myself a larger image size to play with so I didn't need to choose only one focal point.
It's a 4k square, so there's a pretty high level of detail if you click through.
Problems: Some hair clipping. Also, her left hand is strangely floating over the branch instead of making contact... I swear that was not the case when I did the test render. Now that I'm typing this, I think I forgot to save the final keyframes for her fingers... that was probably the problem. The hair, though... that was definitely there and unnoticed.
On a side note, I think I really lucked into a good fur pattern for the tail. The accent strands pick up light in a way that makes it look really dynamic.
It's a 4k square, so there's a pretty high level of detail if you click through.
Problems: Some hair clipping. Also, her left hand is strangely floating over the branch instead of making contact... I swear that was not the case when I did the test render. Now that I'm typing this, I think I forgot to save the final keyframes for her fingers... that was probably the problem. The hair, though... that was definitely there and unnoticed.
On a side note, I think I really lucked into a good fur pattern for the tail. The accent strands pick up light in a way that makes it look really dynamic.
My kids pulled a nasty bug from daycare and got the whole house sick for like three weeks, so yay for not getting anything accomplished.
Anyway, I've seen a lot of talk on denoising applications, so I figured I'd throw in the Maya version for anyone interested. Of course, the AI denoisers like Intel and Optix can be used with anything produced with the Arnold render engine, since... they just require a picture and stuff. but Arnold has its own built-in denoising application (called 'Noice') that calculates things like light and shadow angles, base color, and bump maps during the actual render process as variables that are stored in an EXR file. Then the denoiser takes that information and approximates those variables will also running a standard pixel lookup to find outliers. Here's an example of some tree leaves before and after the process.
Before:
After:
This is cropped out and blown up to 200% just to show details better... I don't usually keep the raw images because they're so much larger than the denoised versions (the additional data points increase the file sizes by about 300%). And the tree trunk is a different layer in the composite image, so there isn't any change there.
Some additional downsides: it takes a lot longer to run Noice than it does to throw something into Optix. With my crappy computer, it takes about 5 minutes per image.
Here's the final composite image before:
And after:
For anyone interested, this is a three-part composite: Leaves, tree, and actor. Reason being that the leaves are really dense and complex, yet also static in all of the image sets for this scene (I have a lot based in the tree and more planned), so I can make it once and reuse it. Each leaf has a transparency map, bump map, and translucency map, and I have sections of the tree with difference translucency colors, leaf size, and some other variables. Basically, it's a huge pain in the ass to render even at the crappy version in the before image, so I'm very thankful for a good denoiser. And since I'm rendering as a raw HDR image file, I can tweak a lot of the light settings in other programs (photoshop for example), so I can change tone and lighting to get different times of day (within reason).
Anyway, I've seen a lot of talk on denoising applications, so I figured I'd throw in the Maya version for anyone interested. Of course, the AI denoisers like Intel and Optix can be used with anything produced with the Arnold render engine, since... they just require a picture and stuff. but Arnold has its own built-in denoising application (called 'Noice') that calculates things like light and shadow angles, base color, and bump maps during the actual render process as variables that are stored in an EXR file. Then the denoiser takes that information and approximates those variables will also running a standard pixel lookup to find outliers. Here's an example of some tree leaves before and after the process.
Before:
After:
This is cropped out and blown up to 200% just to show details better... I don't usually keep the raw images because they're so much larger than the denoised versions (the additional data points increase the file sizes by about 300%). And the tree trunk is a different layer in the composite image, so there isn't any change there.
Some additional downsides: it takes a lot longer to run Noice than it does to throw something into Optix. With my crappy computer, it takes about 5 minutes per image.
Here's the final composite image before:
And after:
For anyone interested, this is a three-part composite: Leaves, tree, and actor. Reason being that the leaves are really dense and complex, yet also static in all of the image sets for this scene (I have a lot based in the tree and more planned), so I can make it once and reuse it. Each leaf has a transparency map, bump map, and translucency map, and I have sections of the tree with difference translucency colors, leaf size, and some other variables. Basically, it's a huge pain in the ass to render even at the crappy version in the before image, so I'm very thankful for a good denoiser. And since I'm rendering as a raw HDR image file, I can tweak a lot of the light settings in other programs (photoshop for example), so I can change tone and lighting to get different times of day (within reason).
- May 3, 2018
- 1,403
- 993
Thanks for sharing!
I've only played with multi-layer compositing a bit, to try and get 3D objects to fit into their background (ie, robot standing on table casts shadows from lights in the room, and has a reflection on the table).
You mention the intel filter. Does Maya have multiple filter, or are you saying that there is a site or a program you can use if you have all the different data types (base color, normals, etc.)? I know that both Daz and Blender 2.80 currently use basic noise filtering. Both the Daz Beta versions and Blender 2.8.1 are experimenting with newer denoisers, both using different (or maybe the same) AI. Blender, which I know more about, plans to use an Intel denoiser based on an AI they developed to handle noisier images. It takes things like normal maps and base colors and all that, and figures out what should be reflected, refracted, shadowed, and filles in the missing light ray data. However, it is not the perfect solution. There was a pretty good
I've only played with multi-layer compositing a bit, to try and get 3D objects to fit into their background (ie, robot standing on table casts shadows from lights in the room, and has a reflection on the table).
You mention the intel filter. Does Maya have multiple filter, or are you saying that there is a site or a program you can use if you have all the different data types (base color, normals, etc.)? I know that both Daz and Blender 2.80 currently use basic noise filtering. Both the Daz Beta versions and Blender 2.8.1 are experimenting with newer denoisers, both using different (or maybe the same) AI. Blender, which I know more about, plans to use an Intel denoiser based on an AI they developed to handle noisier images. It takes things like normal maps and base colors and all that, and figures out what should be reflected, refracted, shadowed, and filles in the missing light ray data. However, it is not the perfect solution. There was a pretty good
You must be registered to see the links
about this, it is not so much gushing over blender like too many other videos, but the main thing is (his name starts with an A) that he talks about flickering that comes from denoisers when trying to make animated scenes, and how Dinsey and Pixar are trying to fix this using either high samples or trying to develop a denoiser that interpolates across 7 frames. Something blender and maybe maya may be a way away from.I find that I do a lot of compositing with the intention of reusing a single background panel with multiple different actors. For this one, I had the leaves render out overnight for many many hours, but the actual changing portions (actor and shadows) only take a few minutes to render individually, then I put them all into photoshop and spit out a final image every few minutes. I'm not sure how true it is, but I feel like this allows me the leeway to make complex lighting and sets with my low-end computer that wouldn't be practical if I were to render them out multiple times.
Last I checked, Intel is standalone and not officially integrated into anything, but Nvidia is pushing their Optix package pretty hard (I'm pretty sure that's what Daz is using, but I haven't checked it specifically). Optix has integration into Maya as a post-processor, but.... honestly, it just doesn't seem to have the finesse that I want from this sort of thing. Solid backgrounds (walls for example) end up with blotches of off-color areas, while the actual lines of the image look pretty nice. It does pretty well with gradients, though, so I think that it would work in Daz, where there aren't a lot of solid colors.
I've seen that one, I like his stuff. Some of it is Blender-specific, but a lot of knowledge from one program can apply to others. Likewise,
Arnold's Noice engine does have some corrections for the flickering that you'll see in single-image post-processing. There's a sequential denoiser option that takes full scenes and processes them against nearest neighbor frames. I haven't had the flickering issues when I've done noise correction on animation loops (yet), so while I can't say that it's amazing and wonderful and the greatest thing in the world, it's so far been acceptable enough for my needs. I hope to have an opportunity to do some cinematic-type animations with the assets I've been making in the event that this game succeeds, and then I'll really have an answer.
Also, to reiterate, it takes a long time relative to all of the other denoising engines making waves right now. Last I checked, Intel is measured in seconds, and Optix just throws out a denoised version of the file after the render is finished (at least in Maya). Then there's Noice that takes an hour to process a 50 frame loop animation. I guess there are going to always be tradeoffs. But for things like the bumps on the leaves and stuff, that just wouldn't be possible without the additional processing time.
Last I checked, Intel is standalone and not officially integrated into anything, but Nvidia is pushing their Optix package pretty hard (I'm pretty sure that's what Daz is using, but I haven't checked it specifically). Optix has integration into Maya as a post-processor, but.... honestly, it just doesn't seem to have the finesse that I want from this sort of thing. Solid backgrounds (walls for example) end up with blotches of off-color areas, while the actual lines of the image look pretty nice. It does pretty well with gradients, though, so I think that it would work in Daz, where there aren't a lot of solid colors.
I've seen that one, I like his stuff. Some of it is Blender-specific, but a lot of knowledge from one program can apply to others. Likewise,
You must be registered to see the links
has a lot of Maya-specific tutorials that are basically just animation tips for 3d artists and could apply fully to Blender (he has a specific one for mixing 3d with outside photos and videos that may interest you).Arnold's Noice engine does have some corrections for the flickering that you'll see in single-image post-processing. There's a sequential denoiser option that takes full scenes and processes them against nearest neighbor frames. I haven't had the flickering issues when I've done noise correction on animation loops (yet), so while I can't say that it's amazing and wonderful and the greatest thing in the world, it's so far been acceptable enough for my needs. I hope to have an opportunity to do some cinematic-type animations with the assets I've been making in the event that this game succeeds, and then I'll really have an answer.
Also, to reiterate, it takes a long time relative to all of the other denoising engines making waves right now. Last I checked, Intel is measured in seconds, and Optix just throws out a denoised version of the file after the render is finished (at least in Maya). Then there's Noice that takes an hour to process a 50 frame loop animation. I guess there are going to always be tradeoffs. But for things like the bumps on the leaves and stuff, that just wouldn't be possible without the additional processing time.
So here's a problem I've been dealing with ever since I started: getting better at stuff.
Posing, expressions, that sort of stuff is normal, right? No one looks at stuff they did last year and wishes they were still as good as they once were. But since I'm still basically discovering features that I didn't use before, my work is getting better in very specific ways, and it makes the stuff I did even a month ago look strange in comparison. So now, I either need to stop making things look better, go back and redo old stuff to match the newer output, or just ignore it and keep going as is.
I mention this because yesterday was a fairly obvious example: feathers. One of the characters is a rooster, which is a pretty simple model that I previously made 'feathery' by use of bump and displacement mapping. It didn't look bad and I always sort of treated it as an afterthought.
See example of old rooster:
And then I started making his 'big scene' moments, where he had to be the main focal point, so I did a little reworking to give him some true texturing with a really simple application of feathers (just very short, thick strands of hair, combed down a bit here and there). I DID NOT THINK IT WOULD MAKE SUCH A DIFFERENCE.
New rooster:
For fun:
Posing, expressions, that sort of stuff is normal, right? No one looks at stuff they did last year and wishes they were still as good as they once were. But since I'm still basically discovering features that I didn't use before, my work is getting better in very specific ways, and it makes the stuff I did even a month ago look strange in comparison. So now, I either need to stop making things look better, go back and redo old stuff to match the newer output, or just ignore it and keep going as is.
I mention this because yesterday was a fairly obvious example: feathers. One of the characters is a rooster, which is a pretty simple model that I previously made 'feathery' by use of bump and displacement mapping. It didn't look bad and I always sort of treated it as an afterthought.
See example of old rooster:
And then I started making his 'big scene' moments, where he had to be the main focal point, so I did a little reworking to give him some true texturing with a really simple application of feathers (just very short, thick strands of hair, combed down a bit here and there). I DID NOT THINK IT WOULD MAKE SUCH A DIFFERENCE.
New rooster:
For fun:
- May 3, 2018
- 1,403
- 993
Actually...
I believe I share in the experience. The reason I hesitate at that first quote because sometime when I explore things (like with what you did with the feathers) I go deep, deep into it, just to prove it as a skill I could master. Eventually I unlock the skill, but because it is so specific, I have to relearn it later on.
recently I have been looking into better character deformation algorithms, and I have been too lazy to spend weeks implementing someone else's
You must be registered to see the links
into code for blender, so I just did some exploration of using physics simulation of the skin, and I have just gotten the most natural, realistic deformation I have seen out of it. It may be a while before I use it or something better, but if I fall out of practice with using the physics simulations and fail to find my old files (like I keep doing), then it will be another one of those one time skills I had, and lost, only to relearn later.great cock by the way
This is also a problem, yeah... getting good at something specific, but then not needing it again for months and needing to relearn. More commonly, though, I find that I open up old files and find that they are 1) way bigger than they need to be, 2) way more complicated than they need to be, and 3) need to have their cameras and lighting changed.I believe I share in the experience. The reason I hesitate at that first quote because sometime when I explore things (like with what you did with the feathers) I go deep, deep into it, just to prove it as a skill I could master. Eventually I unlock the skill, but because it is so specific, I have to relearn it later on.
recently I have been looking into better character deformation algorithms, and I have been too lazy to spend weeks implementing someone else'sYou must be registered to see the linksinto code for blender, so I just did some exploration of using physics simulation of the skin, and I have just gotten the most natural, realistic deformation I have seen out of it. It may be a while before I use it or something better, but if I fall out of practice with using the physics simulations and fail to find my old files (like I keep doing), then it will be another one of those one time skills I had, and lost, only to relearn later.
great cock by the way
Don't have much new stuff... I'm working on simple story elements to flesh out characters. Here's the cliche 'laying down and looking at the sky when a girl finds you and looks down at you from above' scene.
As is always the case with me these days, this is a composite. Background layer, then the nude character layer, then the clothing layer (so that I can tweak the final image when I eventually add user-dictated clothing choices).
Just for the curious, here are the individual layers:
BG:
nude actor:
sweater:
additional clothing option:
And then the composite with the other outfit:
(There's some clipping on her belly that I'll be fixing in photoshop... I didn't want to bother morphing it for the render when a minute with the clone tool was good enough)