ZBrushCentral

Job offered: basic sculpting of faces

Community
#1

Hi there,

We are looking for someone who can create 3D Images from standard photographs of people.

It is NOT necessary to sculpt/create a detailed full 3D image,
but to add a “little depth” to a standard photograph of any person.

That is, parts that are not visible in the photo, such as the back of the head, should not be “sculpted”.

Example:

(pictures removed because the person on the picture changed her mind in regards to me publishing it :cry: )

As far as we know ZBrush, the required task can be done by placing the photo either on a plane or a sphere, and then “wrap” the plane (or sphere) together with the photo so that it somewhat fits the real 3D shape.

Again, we don’t need high details, just the very basic shape of a face.

The photos of any given person will show shoulders, neck, face and hair. Some photos of people may include glasses, beards, mustaches and possibly hats. Format of the photos is JPG, Background of the photos to be left unconsidered.

We believe that this task can be completed in less than 15 minutes with a software such as ZBrush, by any medium skilled 3D Designer.

The result should be made available in OBJ Format with the original photo as texture.

Please let me know whether this is all possible as described (I am not a ZBrush expert) and whether you have some spare time to be able to do a few pictures per day for us.
If you think that you have the time and the skills to do this, we can then discuss how much you will be paid.

Regards

Etienne

#2

Hi etienne,

My names is Joe. I’m a student who is looking for part time work. I enjoy sculpting in zbrush and particularly enjoy sculpting realistic heads and faces. The method you suggested is possible but I find that manipulating a generic head model to fit the face is easier and gives much better results in just a little longer time. I have work posted of my self portrait that is a work in progress. Its in the self portrait section under challenges in this site. You can see my work in my user gallery or you can e-mail me at [email protected]. Thanks.

#3

Removed by request

#4

Hi Joepaddyroe,
I was just about to reply, when I noticed that you posted an example.

your method is nice, and will lead to very good result, but we already see several big disadvantages:

  • the long hair is completely missing and it will require much more work to add the hair.

  • This will become even more complicated if we supply a photo with a hat and maybe some sunglasses…

  • the high detail will require a lot of time, and time is expensive.

We think that it will be absolutely sufficient to create a plane, apply the photo as a picture onto it, and then start pulling out the nose, front, mouth and so on - as a skilled user, you will be done in just a few minutes.

Let me know what you think.

Regards

Etienne

#5

would suggest looking into two products.

Poser ( has a feature called face room, which will do some
if not all of what you want, buying an earlier version than 7
would give you the same feature at a cheaper price)
Face-Gen a small program that will do exactly what you are requesting
and at a smaller cost.

edit: add Face Shop Pro to list for reasons given for others

#6

I’m not really sure how it can be done. I know you can use the texture as a mask and then push and pull points but that can only go so far. The images you have there look like a laser scanner or something? Perhaps if you had a side photo too? Have you seen this being done with zbrush before or did you hear that it could be done? I’ll keep trying.

#7

Hello Etienne,

Quote:
We believe that this task can be completed in less than 15 minutes with a software such as ZBrush, by any medium skilled 3D Designer.
Please let me know whether this is all possible as described (I am not a ZBrush expert)
We think that it will be absolutely sufficient to create a plane, apply the photo as a picture onto it, and then start pulling out the nose,
front, mouth and so on - as a skilled user, you will be done in just a few minutes.

As you mentioned above that you are a non-‘ZBrush expert’, what then leads you to believe such images can be recreated in ‘just a few minutes’ ?
I’m guessing this is more a budget concern, as you mentioned ‘time is expensive’.

Using photographs to extract relief and/or ZDepth information may not be as simple (or indeed as quick) as it at first might appear -
more especially in using a simple plane as a base for organic geometry :slight_smile:

We believe you’d do well to read bicc39’s advice above, or perhaps even engage the services of a 3d-scanning facility for your purposes :slight_smile:

Do you have a budget in mind per model, by the way?

Thanks in advance for your answers.

Chris

#8

Removed by request

#9

Etienne, can you retrieve this point cloud as an obj file?
It could help to transfer that into zbrush

#10

etienne78,

I thought i’d give it a try and this is what I have found

The first image
I took the jpeg and adjusted the levels and saturation to prep it.
Here is the first problem, there is alot of compression artifacting (prehaps a cameraRAW file at a higher resolution 4K would work, prehaps not).

The second image
In Zbrush I used a material which allows for real time rendering of bump maps.
The second problem is that the image hasn’t nearly enough depth data to produce the “relief” style I assume you want.
Note: the tonal range is relatively shallow.

The third image
Here I have attempted to use the bump image to extract what little depth data there was and applied it to a plane consisting of 4million+ polygons.
As you can see the image is rather flat and ‘jaggy’.

Although the time spent in Zbrush to produce the final result was little more than approx 15mins the result is rubbish, due to a ‘knock on’ effect with regard to the quality which started with the original photo.

Even if the orignal photo was a cameraRAW file you would need better lighting conditions, then it may have to be converted to HDRI so as to retain RGB info to possibly work as a 32bit displacement on a polygonal plane? Although I doubt it.
Either way this would be a laborious and time consuming task as you would need to reverse engineer a displacement, which to me sounds near impossible considering displacements are calculated from the height difference of two points, and this is how it works in ZB.

For the best quality out of Zbrush, I would suggest sculpting the figure then generating an alpha from that which consequently can be used to displace a polygonal plane. The actual work might only take 15mins but the prep could take hours/days.

:wink:

#11

Going by their example, it looks like they want more than just what would be considered classical relief sculpting. It looks more like they want a ‘full 3D’ version of the camera view emerging from a plane (vanity wall plaques to give to your granny? :slight_smile: ).

If that is correct, it won’t be possible to get the correct depth info required from just the photo to make this a quick, automated task.

Given that they don’t want too much detail, I’m thinking one of the quickest ways would be to make a couple of generic heads (male/female) and props, grab alphas from these, use the liquify filter in photoshop to shape the alpha to the specific photo (or you could use ZBrush to do this also) and then use the modified alpha to generate 3D or inflate a plane3D through a mask of it.

#12

@bicc39
thanks for your suggestions, however we have tried Facegen, 3DMeNow, Faceworx, Poser and many other tools.
The point is that they all more or less very nicely fit the texture to a standard model, but the modeling itself is not really convincing.

Another big disadvantage is that all of these products, logically, only treat faces, but not glasses, hair, beard, mustaches.

@atwooki/Chris
you are right, I am not an expert! But I have played with Zbrush a little, and I understand that what we need can be done this way - joepaddyroe has proven it - even though he needed more time than I “guessed” :lol:.
We do have 3D Scanners available, but if the person is not there to be scanned, it is difficult to scan the person :lol:

@joepaddyroe
Well done! that’s how we need it. I’ll be contacting you by email.
30-40 minutes may also be acceptable. Maybe you can downsize the quality/detail even more to further reduce time. I’ll be contacting you by email.

@Sebcesoir
Salut Seb, no, unfortunately the point cloud is not available as OBJ, otherwise I would have posted it. however joepaddyroe has posted the result that we actually require.

@Deusex
thanks for trying, but this relief type of thing can be nice if you want to use it for engraving into metal or similar, but it is unfortunately not what we need.

@Bonecradle
You are right that an automated task will not do it.

Unfortunately I do not understand your suggestion, but If it gives a result as shown in the example by joepaddyroe, then it’ll be good…

@everyone
any further suggestions ?

#13

Cool.

#14

etienne78,
Your right, which is why from a single image you won’t get the result or ‘under cuts’ to produce a sculpture in the round.
My motives for trying were purely an exercise in personal R+D within the limitations.
I agree with everyone else that a model is required especially if you want hats, glasses etc…
I’ve supplied some images of what I meant but by no means am I interested in the work, joepaddyroe is clearly on the right track.