ZBrushCentral

Undeletable projection artifacts

Hello there,

I’m currently trying to project the hi-rez version of a dragon body model I created onto my low-res base, but projection results in artifacts I cannot remove with conventional means. Here’s an album of issues I’m mentioning in the post, as the forum doesn’t let me post more than 2 links: Imgur: The magic of the Internet

The low-res base has been sufficiently subdivided, and a layer and a morph target were created before projection.

As I would expect, especially with a model with sections this close together, I have artifacting issues, which I would be more than happy to gradually smooth or morph away and resculpt.

However, on the hi-rez model, the spaces between the dragon scales are quite deep, which isn’t the case on my low-rez base, as manoeuvering into them during retopology in Maya was practically impossible. When I try using the Morph brush to restore the low-res state of the model in those areas, I am left with small holes that collapse deeply inward and refuse to be morphed back no matter what angle I try to brush them away from. Attempting to smooth them causes them to expand. Gradually working from the lowest subdivision upwards has them appear anyway near the highest subdivision levels. My guess was that these are areas that are latching onto the underlying hi-rez surface at the base of the scales within the crevices, which I can’t really reach with the morph brush when these issues happen.

So then I tried adjusting projection distances, but I couldn’t find a value which helps with these spaces without massive harm to the rest of the model.

I tried messing with PA Blur and Projection Shell values too, but all values created other artifacting problems while not really alleviating this one.

The original hi-rez model has each individual scale as its own polygroup, so I tried converting them to polypaint, projecting it onto the low-rez, then converting them into polygroups again on the low rez with the idea of selecting them then projecting area by area, but the polygroup conversion, due to detail difference I suppose, messes up in very small areas and scatters random patches of color between polygroups across the model.

I figured that filling the spaces between the scales more in the original hi-rez mesh would most definitely fix this, but doing this by hand between each scale would clearly be immensely time-consuming, so I set out to try and find a more elegant way to do that. I tried achieving this with Boolean operations, but couldn’t really find a solution with them. I tried exporting to Blender and Maya and using various Shrinkwrap modifiers and values to try and achieve this, but to no avail. I also attempted setting the scales as a collision volume and set the base within to expand via Dynamics, but it seemed the crevices were too thin to make the base “flow” into them nicely.

I still feel like this last approach of trying to fill the spaces between scales somehow would be the cleanest solution, but I’m now completely out of ideas on how to approach that aside from manually clay-builduping/moving the interior of each scale to remove the gaps between them. Any ideas on how to do this, on what I may have missed, or on how to get rid of the artifacts otherwise, would be much appreciated (and thanks for reading the absolute bible of text, but I couldn’t make it shorter). Really at the end of my rope here.

Hello @Tetractys ,

If you have images you’d like to share, just upload them directly to the thread. You can drag and drop them directly in the text window. That way all information being referenced is kept in the the thread, and not in third party links that can break.

You can protect any part of a model that you don’t want to receive projection with masking.

The most common adjustment that needs to be made with Projection is the .dist slider. Lower values here can result in “holes” in the projection. It is rarely necessary to adjust the blur or projection shell.

This movie may be helpful:

Hi Spyndel,

sorry about the images; I simply didn’t want to make the post even larger since the images are a fair few. I’ll keep the advice in mind for next time though.

I had actually run into this very video while searching Google for the solution - I definitely learned some things I didn’t know, but it wasn’t quite what I was looking for in this case. I’m aware of masked projections, but unfortunately, since all the areas in question are little nooks between each scale, masking them properly takes about as long as connecting all of them together with the Move brush, even in these areas that will normally be invisible from any perspective.

This issue was kind of blocking the continuation of my workflow, however, so I’ve already started to trudge through the process of manual corrections/closing the space between scales with the Move brush. Many thanks for your comment either way though!

@Tetractys

fwiw - I’ve had similar issues too with a model of a snake with thousands of scale (small scale) and the snake body (large) and winding around itself, so parts of it are near to each other without touching.

Problem I see is the mix of scales (geometric, not snake!) and/or distances ranging from large to very small. So a single distance (proximity) slider is insufficient. What I’ve done when I get these spikes is undo, wind back the projection distance to a smaller value and re-project. Repeat until no spikes or maybe only a few which can be cleaned up … see below. Then increase the projection distance and project again. The areas that are already projected won’t change much, if at all. The larger scale, more distant parts will then project.

You can also clean up local areas after projection by setting the high res as a history point, smoothing and/or relax smooth the projection in the local area that’s poor, and then clean up with the History Project Brush.

hth

I had a similar problem today with a probably a bit simpler Mesh.
But when projecting my hiRes Head Model to another subtool (head and body) it always gave big artefacts, although the head (from head and body) was a copy of the hiRes head before deleting higher subdivisions.
After many tries I made an experiment.
I just duplicated the hiRes head and projected to the copy, and… Artefacts!
Projecting on to a clone :open_mouth:
So the original head seemed to be “out of order”.
Because when projecting the clone to a “clone of the clone”, it worked!

So I cloned the original head and deleted the original.
From there on it worked again.