Need your feedback

Stensword

So I've been working with references and the results are definitely more realistic than without it, the faces subtool shown below are 500k points to 900k points resolution :

(The models are ranked in the order I made them)

The 3 last one took me around 1 hour each (hair not included) to finish..

Basically I try to sculpt the overall shape till 36k then I start to build the nose and the mouth with 70k, I refine them with 200k then move to the eyes where I jump to around 500k-700k. Finally I refine the facial details + ear and I am done. This is my process.
Can you identify a common (fundamental ?) mistake in those portraits ?

OTHER IMPORTANT QUESTION : I am hesitating between 2 courses, likeness sculpting with Frank Tzeng and the Anatomy of facial expression book, regarding my previous work, which one do you think will be better for my case ? (or even the Understanding the human figure book)

gnoop

in our age  when a nice head model  you can do trough photogrammetry  just asking a person to seat still and hold her/his breath for a 30 sec  shooting like 20 shots around   or with just  lidar phone capabilities  it's IMO kind of  not very competitive skill.    

  But if you need  it to do it  from those two pictures only I would suggest to match the profile first using see through slider on top right in zbrush

Don't forget to set perspective camera to match focal length to the photo.   Don't use ortho view to model over a photo

Eric Chadwick

Hey that's @poopinmymouth

Stensword

gnoop said:

in our age  when a nice head model  you can do trough photogrammetry  just asking a person to seat still and hold her/his breath for a 30 sec  shooting like 20 shots around   or with just  lidar phone capabilities  it's IMO kind of  not very competitive skill.     

Thanks for your reply... Damn, this sort of thing should cost a lot ! The only way I can think of to be able to compete with this sort of stuff would be to cost less and be very very good to finish a perfect likeness model in like 1-2 hours, the adventage with a 3d artist is that you don't need the people at a certain location, you can have a bunch of picture and make it work. But definitely a worthy challenge !

gnoop said:

Don't forget to set perspective camera to match focal length to the photo.   Don't use ortho view to model over a photo

Oops ^^

Stensword

Eric Chadwick said:

Hey that's @poopinmymouth

Who is it ?

Eric Chadwick

The first and last references are Ben Mathis, a long-time member here. http://www.poopinmymouth.com/

gnoop

Stensword said:

Damn, this sort of thing should cost a lot ! The only way I can think of to be able to compete with this sort of stuff would be to cost less and be very very good to finish a perfect likeness model in like 1-2 hours, the adventage with a 3d artist is that you don't need the people at a certain location, you can have a bunch of picture and make it work. But definitely a worthy challenge !

A key thing to likeness is a right proportions and a general scull shape.       You don't need expensive multi camera rigs for that . Just   10-15 photos within 90 deg sector of a face  and  a person seating  more or less still.  The mesh may came noisy   but would still have perfect likeness.     A key  would be just not to spoil it during  clean up and detailing.

As of the "likeness" skill itself  don't be obsessed with it.    I  spent a lot  in art training  doing that  and  ok  I could earn few pennies  drawing portraits in streets.   But it's not 19 century to make a career out of it .

Stensword

As of the "likeness" skill itself  don't be obsessed with it.    I  spent a lot  in art training  doing that  and  ok  I could earn few pennies  drawing portraits in streets.   But it's not 19 century to make a career out of it . 

I want to learn to make likenesses because triple A 3d artist surely know how to make them easily, so If I want to have the triple A level I surely need that level of skill... Also it strengthen your observation muscle and I also believe it's one of the most difficult thing to do in modeling, so if I am able to do that, a lot of other things will be much easier to do...

But yeah down the road the main goal is to be able to make super realistic face it does not really matter if it doesn't resemble by the millimeter the reference I am using. The goal is to be able to create an asian / african / australian.. men or women only by knowing the main landmarks those specific human races displays. But also imagine you are hired in a game studio and they want you to recreate perfectly an actor's face, this skill would be a must.