@circusboy:
It’s a nice sentiment, but photogrammetry usually works best when there are a lot of photos available and from all angles. The photos have to (ideally) be shot at the same time and under the same conditions. He’s working with a small assortment of photos from almost 40 years ago.
@YoZo:
Let me start with a very basic statement.
I hope that you already have prior 3D experience. Not only is character modeling among the most difficult topics in all of CG, likeness modeling is even harder. More than that, you’re looking to create a film, which is a massive commitment of time, energy, and overall skill. If you’re having a hard time wrapping your head around this stuff now, imagine yourself 2-3 years from now when you finish - IF you finish. No disrespect intended. I just know what it’s like. You can’t go from newbie to pro level, enough to do a short film with a CG likeness, overnight.
Having said that, you are going to have to be aware of a few very basic things.
-
Reference material is key. The more photos that you have, the better. Find as many pics as you can of him. As many angles as possible. The key to any likeness is detail. Humans are made up of many small and seemingly insignificant ones. If even one small dent or corner is off, it can sometimes break the illusion.
-
While it’s find to start with symmetry on, humans are NOT symmetrical. Even the most scientifically beautiful person isn’t 100% symmetrical. Don’t believe it? Take any face front picture and flip it. You’ll see how things tilt or features shift when you do. Better yet. Chop a picture in half and then fill the other half with the mirror version. Looks weird, right? People aren’t robots. CG creations are typically symmetrical and that’s fine… in some cases. Where likeness are concerned, however, symmetry is your enemy.
-
Work progressively. Start with a simple base mesh, something that is vaguely human shaped. Build up the features slowly. Be perfectly aware of the shapes, placement of features, and overall proportions. Only add new detail as necessary. Again, mind the imperfections that ultimately make us more believable.
-
Work with clean topology. If the face is going to animate in any way, be it a simple pose or in full form, make sure to retopo you initial sculpt so that your edge/face loops respect the core muscle groups. If you don’t, you might end up with uneven detail and, worse, deformation that doesn’t look right.
-
When texturing, keep in mind that human skin isn’t one flat image. Skin is made up of multiple layers. Each layer respects light in a different way. The way it absorbs or reflects light can also depend on various conditions. Sweat. Oil. Small “peach fuzz” hair. Pores. Good skin comes down to three basic elements: A good skin shader with subsurface scattering. A multi-layered approach to texturing. Quality lighting/rendering. You can add in other elements for more complexity, but those three aren’t up for debate.
-
Even with the best modeling, shading, lighting, and so forth, it STILL might not look believable. Be aware that the “uncanny valley” can be a bit of a bitch. The closer we aim for human likeness, the more we’re likely to find flaws or be freaked out by it. The human brain is designed to recognize other humans. When something is even slightly off, we’ll notice it and it’ll make us uncomfortable. That’s not to say that it can’t be done though. The film industry does it all of the time. They used CG doubles in movies such as “Logan” and “Iron Man 3” in the past. Unless you had a trained eye and were looking for the double, you might not have noticed. It CAN be done, but the level of skill required to do it is high. The time required to get it to look just right is high too.
Honestly? If you’re even remotely new to this sort of thing, good intentions or not, you’ll probably fail. That’s not a dig at you or how skilled you may be. I’m just saying that some topics, especially one such as this, are (almost) about as advanced as you can get. Artists like Hossein Diba (www.artstation.com/hossein_diba) have spent years honing their craft. It’s not easy.
Going forward, before you attempt ANY likeness, just keep on practicing some basic heads and poses. Get some books on anatomy such as “Cyclopedia Anatomicae” and “Anatomy for 3D Artists”. Having those references for your grandfather are key, but there are times when you’ll have to fill the gap in your knowledge. Keeping a good anatomy book on hand will save your butt. Furthermore, invest in a good hand mirror or take some selfies. Because you don’t have a lot of pictures of your grandfather, you may have to use yourself as a reference point for certain poses or expressions. That CAN tire you out at some point, however.
Spend about $20 and buy “Facial Expressions: A Visual Reference” by Mark Simon. The book is about 15 years old, but it’s invaluable. All it is is just 250+ pages of photographs of people making lots of weird and not so weird faces. Young. Old. Man. Woman. Black. White. All sorts of people. There’s a version of the book focused only on kids too. Both book are older, but still pretty easy to find on Amazon. Both are keepers. (I shot some similar poses of myself many years ago. I keep them on tap for when I need quick reference material.)
Everybody’s version of happy or sad is different. Just be aware of that. None of these may accurately reflect your grandfather’s expressions. However, at the very least, they can act as a rough blueprint to help you on your way. Combined with the few photos that you have, that should help.
I wish you the best. I hope that you succeed. Just be aware that you’re going to have to be at the top of your game. There is a LOT of stuff that you’ll need to know going in. Good intentions or not, it’ll be an uphill battle. Tons of work ahead of you. Best of luck.