Texturing inside surface of voronoi fracture object
MaverickMongoose last edited by Dr. Sassi
Hi Dr. Sassi,
I'm trying to create a vegetable chopping scene using the Voronoi Fracture.
I've got the pieces slicing off however a cant get the inside faces to correctly show the texture.
I've tried flat projection but as the vegetable is thinner on the ends this doesnt work, then I tried to make a copy of the voronoi and give all the inside faces a correct UVW, which works well on the copy.
But is there a way to then transfer the correct UVs back to the voronoi?
I've attached my scene with the editable mesh correctly showing the inside faces.
Thanks for you help as always!
Typically, one can harvest UV from an editable object "copy". In the case of the Voronoi Fracture, this is unstable for various reasons. Which I think is not a production-ready idea; hence, I avoid it. Yes, it might work here and there.
I have changed the texture slightly and converted it to ACEScg, as I am not a big fan of sRGB (too many flavors and different gamma settings). But use what you need.
I harvested the model and used the Fracture object. I hope that works for you.
BTW, the "source" can be just a Matrix Object, or as I have used here, just a spline.
Yes, I have set up the file, as something was off.
Perhaps rotate some UV for the cut area, to avoid repetions.
Enjoy the rest of your weekend.
Was hoping there might be a way to keep things procedural, oh well the fracture object is the way to go then.
Regarding ACEScg, I think I set my file up correctly for an ACES workflow, but I created the images in photoshop so I would have thought setting them to sRGB (or AUTO) in Redshift would be the way to go - am I better off setting any color image to ACEScg instead?
Thanks for asking, MaverickMongoose,
Where ever I can I stick to the best quality I can get, and if possible, I love ACES 2065-1, especially for HDRI.
Here are my two cents, and it is up to anyone to stay in integer (to make is easier and simpler to handle) or do it with current workflows.
Since we entered the Linear Light idea of rendering around 15 years ago, I found that the idea of a super limited color space (sRGB/Rec. 709) would be out of the precision we can use with floating point.
The main problem with these formats (integer [8 bits/channel], limited in dynamic range and gamut) is based on their production.
First, the camera has a sensor response curve; the dynamic that current sensors have is beyond what sRGB can hold.
How that is packed into that small format might create distortion.
Is it just an S Curve, Global tone-mapping, etc.?
The rule for HDR is typically a Graycard and Diffuse White. The tone values should be linearly oriented on these two points.
This is typically not the case with most of these texture maps. If one has textures done along the same procedure, the result is more likely to look more plausible. If that is all over the place, well, guess what happens.
It can fill a book or a few more to go deeper, but in a nutshell, Rec. 709 was created around three decades ago, and a few years later, sRGB used the primaries and some similarities of it. It was just enough for the phosphor quality of the CRTs back in the day, which is no longer the limit.
The sRGB colorspace was developed as a delivery format for the web. It is small and with 8bit per channel small in file format. From my point of view, it has no place in a professional pipeline. Let me explain:
It is easy to say that it doesn't allow for the quality we aim for with ACES, DCI P3, or Rec.2020, especially not HDR; the colors we can reproduce today will be clipped in sRGB.
You might think it can hold most reflective colors, like Pointer's Gamut, but that is only a marginal argument. Pretty much all public traffic lights and police car lights are by default or more saturated to stand out. Hence, the weird colors they show in many movies, as many cameras couldn't reproduce them many years ago. In Cinema 4D, setting those up is not a big problem; any limited texture will look pale.
Think of the object you have in your project. It has a lot of green tones, but those green tones are in a small area of the whole gamut. Even with millions of colors, monochrome areas (like a blue sky gradient) can easily lead to banding. To get future-proof material, the bit depth and the gamut must be larger than the results on the end.
Yes, I worked a little bit on the textures; otherwise, just converting a JPG to something bigger will have no effect.
If a screen can't represent it, it might not show at all, but as I have written often, screens get better, and theater laser projectors (high-end) can reproduce Rec. 2020 already. In other words, what is not in the texture can't be on the screen. If conent based on sRGB textures is shown between full Gamut based (ACES/HDR) materil, the sRGB material might look poor.
Current digital cameras have much more to offer than sRGB and 8bit/channel. To cut it down creates no advantage.
There are many sRGB ideas available (with various gamma curves and values; there are even officially different versions, and I like to stay away from them), and over the past two years, sRGB and ACES seem to create the greatest problems, especially when one follows some advice from the web. Auto is not always working properly, like raw is not always coded raw. Same problem. Hence, my take is to go more with native ACES formats from the start. (Yes, I'm aware that there are many counter ideas available)
To have a texture collection that is built with precision, where the color temperature and tint are known, and how the white point was set up, will pay off in the future. I encourage going full float for a long time, and yes, even today, people stay with decades-old formats. Pick what you like, of course.
Wow thanks for the detailed response!
But setting a normal old image (jpg, png etc) to ACEScg doesnt give it a higher gamet of colors does it? Wouldn't I need to create the image as ACEScg from the start or convert it in Photoshop before briging it into C4D?
Dr. Sassi last edited by
Throwing a jpg into an OpenEXR container alone does nothing.
As I wrote, I colored and wrapped it, and I did not want to lose the color. As you can see in your original image, it had a hard border and an unnatural green frame. Since every cut had a different shape and size, I had to find something that worked: the hideous "one size fits all". In this way, the edge of the cut was covered no matter what. As you can see in the UV, all cuts have been scaled to fit the image. Something the Voronoi Fracture would be able to do. So, lots of tiny decisions to make it work.
Back to color, and I'm glad you asked: My main point is in the controlled conversion, with a check immediately to see if it was OK after saving. I have set the Color engine in Photoshop to ACES 2065-1 (AP-0), so the conversation was made in the widest space Ps can offer for that (Well, CIE_RGB might be bigger, but that is bean-counting to compare the two). With this, I'm confident in my conversion.
One of the reasons I do not like to use sRGB inside the pipeline is that one never knows how the material was treaded if not done on one's own from raw, with a gray card, and light meter, perhaps even a MacBeth chart and Colorimeter (my trusty Sekonic C800 for example). What Gamma was used, along the books or just a simple formula? Which one was it, and which version or year was it done, older sRGB can be based on gamma 1.8. Then the sRGB could be a Tonmapped from ACES, which would work perhaps with an OutPut ODT used as IDT but which otherwise likes to explode any "normal" sRGB to something it never was. I could go on and on about this. In short, I don't like to use low-level stuff when spending time on something. It is not worth it. Similar to putting a jpg into Open EXR, one can also store a lot of nonsense (colorwise) into an sRGB, but there it is worse because it demands to have everything small from the get-go.
I have some plans, and since so much conflicting data is distributed on the web, I would like to go a bit broader and deeper into it next year.
My background in Color Correction Grading started early, but my first few longer courses were in 2007 over at FXPHD. Then, after the 2012 release of ACES, I took three ten-part Color Science courses with Dr. Poynton, who does a lot for SMPTE, for example.
Since then, I have taken many courses, and I even took a six-month semester this year at The Summer Academy for Professional Colorists here in Hollywood. I am finished with the highest level 3.
Since Aces 1.3 is out, I took four courses about ACES and will continue to do so. I'm happy to share book suggestions as well.
Perhaps that explains why I'm picky with color and have a different idea of what I use to get my results when I go into a color grading session.
Learning never stops, and developing an eye for color is perhaps more a journey than a short little adventure.
In short, to put perhaps weeks into a project and then notice to have perhaps some colors off or other problems seems to me unacceptable.
With HDR and its color spaces and dynamics, we need to adapt; otherwise, clients will go to artists who can provide the best quality.
There is a learning curve, and I hope to get something established next year that helps me feel at ease.
Sorry for the long text, but since I write in a forum, this is also the space to address it more broadly.
Wow, you've got some very impressive color credentials there!
In that case I think I'll try to adopt your approach in future, cheers.
Thanks again for the very detailed explanation!
Thanks a lot, MaverickMongoose; very kind of you.
Sorry for the bragging, but I like to share the "source" of my knowledge.
If you would like some book suggestions, my favorite Color Science book is from Micheal S. Tooms, but the Mark D. Fairchild book has a huge value. It's not a "light" read, but after working through those, things look different. Pioneers Of Colorscience is a Who is who of this field.
My suggestion is to read Tooms first. If something is unclear, check it in Fairchild's Book; if it needs to be deeper, The Pioneers of Color Science will hopefully close the gap.
Then read the second book, check the third to close new gaps, and go through the third one to finish. This will set a good base.
Sadly, these books are not free.