Trying to improve likeness sculpting
Just wanted to write down my thoughts and what I’ve learn so far
I don’t really feel like there’s many good tutorials on how to capture a likeness probably cause the main talent it requires is observation.
At this stage I’m really trying to come up with a step by step process that doesn’t have me aimlessly sculpting, I found it to take a shocking amount of time, and once you capture a likeness it can be really fragile and easy to lose if the wrong adjustments are made, so it’s constant back and forth evaluation.
What I’ve found -
(Obviously learning anatomy of the skull and face firstly at a basic level)
- Reference images, profile, under, above, 3/4, a HERO image front/profile that’ll be your main reference throughout. It’s really important all the images LOOK like them/look like how you want to capture them, people can look so different from one photo to the next
- You really need to identify the shape of the head first, eg Heart, Square, etc. and the ANGLES of the face eg recessed jaw, like the direction of the shapes and how far back or forth they sit from each other
- Gestalt/squint test: maybe take the hero image and blur it. This is the most fundamental part of the entire process. To be able to create overall the head shape, placing the outline of the features at the correct distance and making sure the head is angled correctly all at a very low poly rough level. If Ive sculpted for 5 hours and have already moved on to details I'm doing it wrong, a long time needs to be spent at this rough stage, I try to match shadow and highlights of photos using basic material shader to match the correct volumes of the face, if a light is constantly hitting the face at certain spots that can be important information
- Gestalt - polypaint rough eyelash, brow and hairline checking against references
- When happy move onto capturing the more specific shapes of the individual features, the forehead, cheeks and jawline all should have been 90% done at the gestalt stage. This is for the shape of the eyes, lips, nostrils, I usually go first to the brows as they are the largest and most complex shape with how they connect to the nose bridge, upper eye lids and upper cheek bone. For individual features I found cutting out your references in photoshop to see how they sit on a screenshot of your model can be helpful. Seeing the lips, nose, eye shape how you want to look on your model at the correct size is less work translating it from a reference and allows you check likeness and make sure you’re going in the right direction
Replies
- beeing able to "see" the difference by precise observations and "only" make hold of this to have this in your visual library/knowledge base/memory (meaning brain and not something on a computer) so you know what to do after this practice
- to refine your skills so you do not even have to think about this anymore and especially with "real" manual work (with your hands on some "stuff" ) you simple "feel it" and this is the "secret" to mastery
I would love to be able to do some things after dozen of hours of work when some people are able to do more than me in the first minutes they start their project. ( And sometimes it's funny when people think i can do awesome things when i think this was easyMy favourite 3d artists are so exceptional at likeness, it's shocking how many years it must take to be as good and frequently good as artists like Hossein Diba and Hadi Karimi. I found some people are much easier to capture than others so maybe that can be a factor.
There's a lot of amazing 3d artists that I see use the same styled generic anime waifu face, this obviously lets them spend so much more time on the rest of the character, but I can't look at it with same respect as I do those that do amazing hyper realistic likenesses.
A lot of tutorials suggest using spotlight in zbrush to align your mesh to photos, I remember trying it in the past and don't recall it working too well, obviously photos are 2d, they're not at a perfect straight on angle, they have perspective etc, and I find I can't really even see what I'm doing to my mesh with this photo pasted over it.
Trying to make precise measurement like they do for wax works seems interesting, but probably reliant on really solid references.
The whole spotlight method interests me cause it'd be great if it was just as simple as align your mesh to a good front and side reference photo.
Breaking down the face before you sculpt it such as going into photoshop overlaying different reference seeing how they match up, measuring/estimating overall distance from chin to mouth, nose to eyes, forehead to hairline, and doing silhouettes and outline drawings of the head/features to understand the shapes before diving right in to sculpting has helped me
https://duckduckgo.com/?q=focal+length+affects+facial+features&iax=images&ia=images&atb=v340-1
..so the "traditional" way of practicing portraits by using a "live reference" may be some lecture one may have to go through to get the experience to make this.. (and for some celebs one almost does not really know how they look from some angles or one would not recognize them).
(Telling this even when i suck at drawing or sculpting at all
I was looking at Amelia Rowcrofts process videos, she has her own likeness tutorials on her website. I noticed how little she adjusts the features back and forth after the initial foundation, so her method of measuring the face must be really solid with how little is being adjusted at later stages
eg. from this stage to this, the features seemingly aren't being adjusted
another example, see how from the 4th pic every remains in place, I wonder how much experimentation she does like if she tested the measurements out before this final sculpt