Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.Daz 3D is part of
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
So, I played a bit with making leaves. I did not yet create the full set of texture maps (just one, no separate bump or transparency or so. And my leaves are not as complicated as those by you, Rashad. But here are the results so far. It was supposed to be a bit like a Magnolia, because I would like to create the tree with flowers (the flowers will be a real challenge, I guess). The form of the tree is not very Magnolia now, but still wanted to show it.
Lights by Rashad (EDGLS) and sky also by him (including the cumulus cloud layer). I upped the bounce light dome a bit to get a bit more light on the shadow regions. Also made the colour of that ligth a bit greenish. The grass is from Rashad's grass clump. Tree created in NgPlant, leaves in Hexagon, textured in PS Elements and added as G-mesh in NgPlant. I did bump my head against a memory issue when exporting from NgPlant, but after lowering the number of leaves, it worked fine.
To show the leaves in a little more detail, I made a close-up as well.
Hansmar,
Looks like you're getting more comfortable with ngPlant! I mentioned it in the other thread but I think the diffuse setting on the grass is a bit too high, the grass appears a tad "radioactive." Cut it down a bit and I think it will all fall into place. As you know I love that sky and those clouds so never any complaint from me there. Nice progress!
Rashad: I agree: too much light in the grass. Even more than in the other scene.
Y'now, I once thought that Rashad's exotic, palm-like, beach-like works were over-cooked, but, I then thought, that as I don't have any experience of tropical environments (the heat, the sounds, the people...etc.,), I was open to understanding.
I love the colour of your grass, Hansmar - personally, I like psychedelic colours (I'm biased), so, while it sounds odd, inexperienced as such, I love this scene.
Jay
Thanks, Jay. Of course, colours and light do not have to be realistic; that is one of the fun things about Bryce!
Heya everyone! Wanted to post a couple shots of a recent scenario made entirely with vegetation from ngPlant. Rendered in Octane. As always feedback is greatly appreciated. Thanks for your time!
Thirsty Buck
https://www.daz3d.com/gallery/#images/1045381
Taking the Kids Out for a Swim
https://www.daz3d.com/gallery/#images/1045371
Hi Rashad!
Such a scene can still be rendered in Bryce with a render time of 1 hour. No transparent foliage, just Ambience 5 to 10. Render mode Premium, TA, 36 - 64 rpp with Bryce Sky HDRI lighting.
Rashad - nice to see you here. Trees, vegetation and PoV look very nice. The foliage looks flat, I'd say its pure ambience if it weren't you that made the renders
. But it is difficult to get it right with a path tracer. I experimented with Octane long ago but prefer the results obtained with a ray tracer, even if it takes longer to render.
SlepAlex,
I agree Bryce could render the scene, but it would take more than 1 hour. This is due to the geometric complexityand TA. But the choice for Octane was more practical. I wanted to play around with a high leaf density, and to do that I needed translucency to be enabled. But there was another consideration. Since I needed to fit everything into Vram I needed to use billboard style leaves, so the leaves in many cases are alpha mapped, as well as translucent at the same time, a state of being which is currently forbidden in the the Bryce Materials Lab. For me the lack of translucency isn't just an issue of the appearence of the leaves themselves, it also affects everything beneath the canopy which is supposed to receive light which has travelled through those leaves. Cutting off that input means replacing it with siomething else, likely more ambient. It woudl require adding a touch of ambient most most every surface, not just the leaves, but the grasses, stones, tree trunks, and most every other item which are beneath the canopy. Not impossible, just more work is all.
My dream is that Bryce does get an update soon and is given direct access to one of the tested and true pathtracers at some poitn very soon. This along with 64bit support. Even without those things I will continue using Bryce. I'm just waiting to build my new rig. the previous dual Xeon system has served me well and was a big leap forward when I first started using it over 10 years ago....yep that long!
Horo,
Yes, we do hates the misuse of that there ambience! Fortunately, I dont often think of it too much anymore, and when I do it's becaue I miss having it. And surely ambience has its place. Ever the smarty pants and spot on observations; Yes, Horo, yes I'd say there's a slight chance I might have over-expressed the translucency on the leaves a bit. And it doesn't help that the sun altitude angle is fairly high, so its a little difficult to discern which leaves are glowing because they are lit directly by the sun vs those which are brightened due to translucency effects. Yes, the darned result look likes flat ass ambient. Go figure. Unbiased and all that jazz...
Rashad
"I needed to use billboard style leaves, so the leaves in many cases are alpha mapped, as well as translucent at the same time, a state of being which is currently forbidden in the the Bryce Materials Lab."
That's almost right, though it is possible to cheat in a bit of translucency. It's difficult and not always good enough, but better than ambience.
Unbiased sounds great in theory but since each unbiased render engine renders a material under the same light conditions differently I wonder what "unbiased" actually means. To be realistic, every render method is a cheat because there are so many things that cannot be accurately simulated (atmosphere for example). There is, of course, nothing wrong with cheating as long as the result looks pleasing to the artist (never mind what others think).
I'm not a fan of path tracing because it doesn't consider what's behind the camera, and that - of course - is a big part of the secret why it renders much faster. But you need specular maps, reflection maps and what nots to add to a material or shader for the object that the renderer doesn't provide because it never looks behind. It's good for games and animation where render time must be very short. Bryce has two render engines built in and the material looks the same, whether you render regular or with TA. Now some path tracing renderers start to include partly ray tracing and they brag about it. As far as stills are concerned, render speed is not much of a concern. You work for hours, even days, for a scene and it ought not to concern you whether the render finishes in 10 minutes or 10 hours. For animated scenes, this is - of course - not acceptable.
I'm not certain this part is correct regarding looking behind the camera. Path Tracers do take into account all surfaces in the scene, even those behind the camera, this is part of their unbiased definition. Path tracing is actually super slow, its only due to GPU acceleration that applications such as Octane can do so much so quickly. Otherwise we'd all still be using raytracers exclusively.
@ Rashad Carter, thats true, it is exactly the same with blender and the cycles engine as well :)
Rashad - I have that idea because I watched a video by NVIDEA a few months ago where they bragged they get the speed with their path tracing because they only consider what the camera sees. And because Iray shaders need specular and reflection maps. I may have got it wrong, but can't find the video anymore.
I think you may be reffering to RTX ...ray tracing, by Nvidia
Which is realtime
Now here is something i wanted to mention that might be useful (or not) for those of you who are doing realistic landscapes and using external tools to produce trees for import into Bryce.
I accidentally found this tool called tree-gen. It's a plugin for Blender 2.8+ (which is also free and can export to different formats). What caught my attention, is that author of tree-gen left a good impression on me due to fact that he studied subject of virtual botanics simulations deeply and actually it was a project for dissertation. Also terms of use are pretty good for enthusiastic hobbyism usage:
So i decided to try it and see if maybe it could be useful for fellow Brycers as well. I could say that tree-gen is most close to Arbaro in how it functions, and i wouldn't say it has much more functional, but may have a couple of tricks, like very thin n+1 branches or tropism settings. Documentation and download link itself is here: https://github.com/friggog/tree-gen/wiki
But the real question, would it be really useful? That's something i can't realy determine as i lack experience of using trees in brycing yet. Maybe tree-gen is just slightly more powerful than Arbaro or NgPlant, but also it produces relatively high-poly results which will eat memory even more than "pro" vegetation solutions? Well, during converting from bezier curves into mesh tree-gen allows to choose between LOD levels (lesser polycount versions). However, those polygonage reduction algorythmes aren't perfect, and lowpoly settings deforms geometry too much, mostly leaves. Also it really depends on a tree species, whether it's a light or heavy model. I think the highest polycount numbers i got when loaded Weeping Willow, it was around 15 millions. On average, though, and with mixture of different LODs, trees are around 600k - 1,5m maybe. Perhaps would be useful in foreground of scene.
As i said, i don't really have much experience with vegetation in Bryce currently, so when i imported mesh to see what i'm getting, i felt questionable whether it was worth the hassle, and render times scared me even on regular render.
So i think i should ask bryce vegetation veterans to try and see if it's a nice addition for list of freebie tools, or it doesn't fit much our workflow.
I took screenshots (previews, open fullsize in tab) of presets that came with tree-gen so you get initial impression.
Acer:
Apple:
Balsam fir:
Bamboo:
Black oak:
Black tupelo:
Cambridge oak:
Douglas fir:
European larch:
Fan palm:
Hill cherry:
Lombardy poplar:
Palm:
Quaking aspen:
Sassafras:
Silver birch:
Small pine:
Sphere tree:
Weeping willow:
I guess not many of us are used or want to deal with Blender, so i decided to upload some mesh exports from these presets so you can try yourself without dealing with plugin itself. In case if you like those, i will post a simple step-by-step instruction how to use that plugin and export results. However, i must admit that i had unstable experience with it, having random crashes here and there. Might be because tree-gen is still version 0.2 and development is not finished. Ok, here are the OBJ exports i prepared to you (i tried to choose lessen polycount for trunk, but kept highpoly leaves to not destruct their shape). In Bryce just ungroup them to select trunk or leaves layer and assign mats.
Palm - 24k polys - 2 mb obj - 500 kb download
- https://files.fm/f/swbwghppy
Bamboo - 395k polys - 35 mb obj - 7 mb download
- https://files.fm/f/frvycypgu
Hill Cherry - 463k polys - 42 mb obj - 8 mb download
- https://files.fm/f/ph3nsfcjt
Small Pine - 684k polys - 66 mb obj - 14 mb download
- https://files.fm/f/aq4684z6c
Cambridge Oak - 2,6m polys - 220 mb obj - 44 mb download
- https://files.fm/f/vetnbydrp
So, that's it. Looking forward your thoughts, tests and verdict.
Thanks for the downloads.
those are much nicer than the trees I made (using a Blender plugin I think was that one)
akmerlow - thank you for the files. I don't think I'll go the Blender way but experimenting with the five trees may be interesting.
If you would need others from presets, or have some suggestions for custom setting (see documentation i linked), just tell and i will export again.
this couldn't be the plugin I had as that didn't have presets
I grabbed this one now, thanks
I have so many Blender plugins some better than others
akmerlow : thank you for the downloads.
akmerlow - thanks again for the tree objects from TreeGen.
All - I imported the 5 trees, gave the wood and foliage each a simple material and saved them into a new Object Library. I measured the memory usage with the Process Monitor looking at Bryce only. Here is the summary for all trees in MB (Import / Object Library size increase / Loaded from Obj Lib):
Bamboo 86 / 45 / 80
Cambridge Oak 515 / 304 / 510
Hill Cherry 96 / 50 / 91
Palm 13 / 83 / 142
Small Pine 148 / 83 / 142
The object library size with all five trees is 485 MB.
All trees consist of the wood (Branches) and foliage (Leaves). If ungrouped and both selected, they can be instanced, then grouped again, and the instances use almost no additional memory.
You are welcome, and thanks for memory measuring tests.
Maybe for heavyweight ones (like Cambridge Oak) would be more practical to use 3d scanned trees like C-ram did (unless i misremember) instead. 510 mb seems pretty expensive for procedurally generated trees exports, depending on what else we have in scene. Do you consider these trees being visually more appealing than arbaro counterparts?
By the way, Hill Cherry should have not only "foliage", but "blossom" (which percentage is set in tree-gen). At least Apple preset was like that.
-
And to those who would tryout tree-gen yourself in blender, i should wary you about memory leaks. When loading presets for screenshots and converting yesterday, i ended up with blender taking around 8-10 gb ram. Solution for that is easy: before next generation of another tree, start a new scene (that would drop ram usage). Also keeps crashing randomly for me during generation.
Hill Cherry has only Blossom, no foliage and I gave the blossoms a leaf material. But this is not very good. As far as comparing Arbaro with TreeGen trees, I haven't made comparisons. Though the Cambridge Oak is very heavy, you only need it once because instancing uses almost no memory and you can resize and rotate it at no cost, only the materials cannot be changed on an instance.
Akmerlow, thanks for creating this and bringing us up to speed.
Horo, would you be willing to share the Object library with us, e.g. via your website? Of course, we could make our own, but I am a very lazy person
Hansmar - I thought about this too. The 485 MB Library has a zip size of 250 MB. You may have to modify the materials, all have the same and for the Small Pine it is particularly unfit.
https://horo.ch/guestxchg/TreeGen.zip
Thanks Horo for the file, much appreciated.
At first glance, judging by the screenshots, these trees are no different from Arbaro. Here is a large .OBJ file of a tree in Arbaro I have 33.8 MB. The scene file with this tree and procedural textures weighs 32.2 MB. The tree contains 581073 polygons.
Akmerlow's Cambridge Oak is unacceptably large for Bryce. Only the OBJ file took several minutes to load. 2.6 million polygons, OBJ weighs 217 MB, scene file weighs 149 MB. The tree looks unpresentable. Conclusion: this is unacceptable in general, for Bryce in particular.
It makes no sense to use trees over 500,000 polygons. Optimally 200,000 - 300,000 polygons, given the fact that the scene should contain several trees, bushes, grass and other objects, and at the same time excellent picture quality.
Nice examples Slepalex, thanks for sharing