Progress on Photorealistic Garments
Hey guys,
So a couple of months ago I stated a project (along with some discussions on the forums here) on dealing with how we could make photorealistic garments from an image of a fabric taken by photograph. This is using MD - DAZ workflow.
Here's the latest update and I would appreciate all comments necessary.
Some issues on the top of my head are:
1. The shadowing doesn't look realistic enough. I am using the basic options in DAZ for Shadowing and Lighting (Sun-Sky Mode)
2. The garment drapes well, but doesn't feel like the actual fabric when draped. It is actually a chiffon dress but in this picture it feels a bit too stiff to be a chiffon dress. I do not have any pictures for reference but i am sure one can imagine what a chiffon dress looks like in this style.
I am not sure if the fabric properties in MD translate well into DAZ, or whether they even translate well in MD itself vs. real life.
So if anyone has any feedback or suggestions please do share, i would really appreciate it.
thank you !

Comments
Are yo sure you mean chiifon which is usually a shear fabric, and tends to be used for more flouncy/floaty sorts of garments and either layered or lined. The dress you have there I'd have thought would be cotton or a soft wool possibly silk.
didn't know what the fabric looked like so looked it up well it doesn't look like that fabric you were going for but other than that it looks great really realistic the entire ensemble if this is a product your planning on selling think you'll do well and you have some good fitting movement sitting morphs for it of course
What makes cloth photorealistic is the details - hems, seams, buttons, etcetera. Just textures and realitic draping can't do everything.
And wrinkles...not just the major ones, but, especially, the little ones.
I've also been working on this. If you're just importing OBJ from MD, the mesh should be exactly the same, so there shouldn't be any problems translating into DAZ.
I've also been wondering about the "photo-realism" of the fabric presets in MD. I think maybe working off a reference would yield the best results? You can endlessly tweak the fabric settings in MD to get something you're happy with.
THIS is why "dynamic cloth" is really difficult for uses like DS. There are many "draping" methods that look good "in motion" that actually look TERRIBLE when rendered still. Standard "spring and collision" methods look good while moving but fail miserably when rendered up close and the details are shown. Hems and seams modify how cloth hangs *GREATLY*. A hem can increase the thickness by more 3x and a sewn seam can add 4x or more depending on what material(s) are joined and the stitch used. This changes the math used dramatically. A long seam will completely change how a piece of clothing hangs, bends and reacts. Want to really throw in a mess: a buttoned coat/jacket front. 2 thick hems + seam + stitching + buttons. Lots of math there folks.
This is why I am unimpressed with some of the latest demos that I've seen shown. Many are based on game algorithms and are not correct for stills.
Kendall
I find if I add subdivision to an MD ouptut in Daz (or any other 3D app) it really adds a zing to the MD obj, and gets it looking closer to what I see in the MD viewport IMHO (regardless of the MD particle distance you set). More detail seems to become visible, especially in the folds.
I have no idea what I am takling abut when it comes to coding and the like but I do have to agree with Kendall on this. It is one thing to have cloth drape like that but what about Buttions, belts and things that are not as flexible as other parts of the cloth.
It's not everything, but, frankly, having easily flowing clothing (particularly if you can dial it down to shift a little rather than a lot) would solve a LOT of clothing problems (like skirts).
Like, I use Dyncreator to fix fairly flowing clothing in a lot of cases. It's a bleepin pain in the bleep. Can easily eat up an hour to get it to do it right. But it often looks a lot better.
Having a faster/easier solution would allow people to try out such things, see if it helps, without spending valuable heaps of time even attempting it.
I mean, heck, at least half the time I go 'eh, this is taking too long and I can't get it to look right' and go back to conforming clothing, having wasted hours.