Wrap texture around a helmet (possible with zbrush?)
polycounter lvl 3
Hello, I've been bashing my head against the wall all night for this and I just can not get it working.
I have to render product shots of helmets for a client, the client has provided a 3D model that their previous 3D artist used. The client design various prints, like a space print, army camo, spongebob face, angrybird face, etc. and I have to place these prints on the helmet and render them out for their webshop. Below is an image of what the helmet looks like and the provided UV map, I've placed on their space textures on the helmet.
As you can see on the image the texture does NOT align at all. Their previous 3D artist told me that he used Zbrush to project the image onto the surface using spotlight. After looking up a few tutorials on that I gave it a go, but with spotlight the texture gets 'stretched' on the edges I had to manually paint the sides.
In the end it did was indeed a bit more seamless than on the image provided below but it looked like a kid went finger painting. And with the space texture I can get away with it looking a bit fadey but with something more tight/neat it's going to be awfull.
I currently have no clue how to get the textures the company makes placed properly on this helmet. Literally every help is greatly appreciated.
I've also included an example of a render the previous artist made, and how neat his texture looks.
Helmet & UV map:
What it's supposed to look like:
I have to render product shots of helmets for a client, the client has provided a 3D model that their previous 3D artist used. The client design various prints, like a space print, army camo, spongebob face, angrybird face, etc. and I have to place these prints on the helmet and render them out for their webshop. Below is an image of what the helmet looks like and the provided UV map, I've placed on their space textures on the helmet.
As you can see on the image the texture does NOT align at all. Their previous 3D artist told me that he used Zbrush to project the image onto the surface using spotlight. After looking up a few tutorials on that I gave it a go, but with spotlight the texture gets 'stretched' on the edges I had to manually paint the sides.
In the end it did was indeed a bit more seamless than on the image provided below but it looked like a kid went finger painting. And with the space texture I can get away with it looking a bit fadey but with something more tight/neat it's going to be awfull.
I currently have no clue how to get the textures the company makes placed properly on this helmet. Literally every help is greatly appreciated.
I've also included an example of a render the previous artist made, and how neat his texture looks.
Helmet & UV map:
What it's supposed to look like:
Replies
OP : You've simply misunderstood what the previous artists explained to you. What was meant was : using projection techniques to progressively project any pattern/texture onto the object. For instance starting from the main angle/point of view, and then carefully working one's way around a model.
The previous artist never implied that the flame texture was projected in one single pass - and there is no reason for you to expect to be able to do that either
Now there's also something else you can do, which is to create a new set of UVs allowing you to work in any way you like, and then projecting the pixel data of the texture you create for them from the new UV set back to the original. But given the nature of your current question this would probably be too ambitious a workflow at this time. So instead simply focus your efforts in learning how to operate the Zbrush texture projection tools (or learn any other tool letting you do that : Mudbox, Mari, 3DCoat).
a quick demo using Tri-Planar in Mari, im pretty sure you can get similar results with designer or Painter
@onionhead_o That's a cool technique, but how much control do you have on the texture positioning and 'flowing'?
I would recommend you to post a clear and concise video showing you running into the specific issue you are having because I feel like there is a fundamental misunderstanding going on here.
EDIT: I know in the video I scaled the model up and down during the painting, I dind't do that on my previous attempts.
Here is a link!
First off it's important to realize that the case of the previous artist (graphic flame pattern originating from the open edge of the helmet) is fundamentally different from yours. The artist clearly used a projection technique to gradually project the graphic pattern along the helmet opening, and then cleaned up the transitions. It very fast and simple to do in that case, and without a doubt you can see how that would work - it's trivial.
Your case (tiling camo patterns and space sky) is completely different. Texture projection is here to literally project textures - not to automagically guess how to fill the empty parts with something perfectly matching the already stamped down parts. There is no "bug" here - it's simply the way texture projection/stencilling works.
Triplanar projection will not help you either, as it will create exactly the same kind of ghosting/overlapping effect.
You have two ways to make this work :
1 - Proceed exactly as you do in this video, and then patch things manually. Now you observed that it is possible, but probably very long. This is true to an extent, but you also have to think non-linearly. Sure, you have a texture (camo, sky) that you want to project and it's "bugging out" where painted parts meet. As you observed you could painfully patch it. But you could also leverage projection painting once more by screengrabbing the model in fullbright mode with your partial paint job applied, send that to Photoshop, fill the patch there using the powerful Photoshop tools, and use that as your new texture to project onto the model at that specific angle. It would would only take a handful of custom patches - it's very powerful workflow.
2 - As mentioned earlier, since this model has been provided as is by your client you probably don't have the liberty to change its UVs. But what you *can* do, is to create a new UV channel (or a new duplicate of the model) with UVs made especially for seamless tiling over that area. Then apply your tiling camo or sky texture to that version (no need for projection anymore), and then convert the texture to work with the old UVs by baking model to model (or UVs to UVs). This is your absolute best bet, and it is doable with any good baking app (3DSMax, Maya, Xnormal can all do that kind of texture conversion). Very useful stuff.
I like the idea of taking it into photoshop, I also COULD paint just the middle part+edges, and then patch the thing up in photoshop. I will definately give that a go.
I can do whatever I want with the model, as long as I can output the product shot they desire. But UV mapping is something I have always avoided like the plague, and now it's on my doorstep and it's knocking hard. I don't want to seem like I need everything to be spoonfed but could you point me in the right direction how this second technique would be set up? Also I don't understand because a few posts earlier you said this, why would it be any different now?:
- Blending Falloff between axis.
- Local rotation X Y Z ( ex. i can control the UV rotation of the image on each axis)
- Global rotation XYZ (this rotates the whole projection)
- Translate XYZ for each axis
- Tiling amount for each axis
But as Pior says, Triplanar does introduce a bit of ghosting so. For a specific pattern like the camo. Tri-planar definitely won't work.
Im gona keep digging at this for a procedural way to achieve this and report back. Because the Technique you use is very tedious and time consuming.
I dont believe there is a universal way for every single pattern out there. For me the deciding factors to which technique I use comes down to How close does it need to match the given design?
Does it need to Align to Center?
Is the Pattern Symmetrical?
That said in the practical cases you are currently struggling with (camo and sky), the distortion introduced by such a new unwrap may be acceptable. It really depends how you do it, how the relax algorithm treats this particular shape, how you hide things, and so on.
There's also a bit of a "cart before the wheels" problem going on here. The definite solution to this problem is pretty clear :
• You need to have two sets of UVs that you can freely convert textures between (the original layout which is great for distortion free-mapping, and and alternate one for global application of complex patterns across the seams, with an acceptable level of distortion).
• And you need a reliable projection workflow allowing you to not only project raw textures, but also project patch elements carefully crafted in photoshop.
But these two points *do* require a solid grasp on UV mapping/unwrapping. So if you do have a little bit of time on your hands before your deadline for this project, your best course of action would be to spend a few days/weeks learning about this and only this, as it's knocking on your door pretty hard indeed. But if you do not have the luxury to take the time to do that, then state it clearly and people will without a doubt chime in to help, probably even helping you preparing the files.
Also ... if the final render is the only thing that really matters, then nothing prevents you from simply photoshopping that texture in after the fact. It goes back to the need to be as clear as possible regarding your constraints : if you ask such a question on a game art forum like here without going into all the specifics of this job, then everyone is going to assume that you need the final model and texture to be looking good from all angles (hence all the projection and UV discussion). BUT if all you need is a good looking static render, nothing prevents you from simply comping in the textures in Photoshop and call it a day.
Lastly : in a way this is all a bit ironic, considering that the product does have a cloth seam there in the first place. Meaning that once produced in the real world, these fabrics with tiling decorative patterns *will* exhibit seams once applied to the helmet, unless they are produced very accurately and especially for this model.
I think you will see that as soon as you have a good handle on these two things (being able to unwrap, and convert textures between different UVs) and also a good grasp on texture projection, the world becomes your oyster.