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
@goofygrape: that looks good! I'll have to try hair-grass more too!
Thanks to you, JStryder
I myself haven't posted in about a billion years. Sadly, or crazily, I am still clunking away at the Volcanic Archipelago. There's just so much one needs to grasp about the instancing lab. Once again I have remodeled all of the ground level vegetation since the last time I uploaded anything. Leaves are more ruffled along the edges and appear more realistic with textures. However, I am unable to use translucency for the leaves, takes too long. In the earlier versions the leaves always looked like cardboard. Now I think I've found an ambience mapping trick to replace the translucency without looking at all like ambient, but you can tell me. I think lighting is improved as well but it's hard to tell. Any feedback is appreciated. Thanks.
For comparison I have uploaded a few examples of the stages this project has gone through over the years.
1. April / 2009. Made with Bryce 6 before the advent of instancing. This was as far as I could push Bryce at the time. Trees have buried trunks revealing only the branches and leaves which give the appearance of multiple trees from a single tree.
2. April / 2011. This was the first attempt in Bryce 7 with instancing. The vegetation would look much better if I had used translucency on the leaves, but alas, it was too render intensive. For me the leaves look of cardboard. This was also the first time I had the test what would eventually become my EGDLS (EarthGlow Dome Light Strategy) which is my go to for all my outdoor lighting for the last couple of years. It is very flexible but relies on the new Dome Light feature in Bryce 7. Anyhow, in this shot the lighting suggests evening to me even though it is plenty bright as I was going for full daytime. Also the flat water plane wasn't convincing to me for ocean water that is supposed to be quite active. Feedback was that foam was missing from the banks, which is true. Notice there are many gaps and holes inthe population along the ground. Grass is in little patches, not at all a carpeting as I would like.
3. January / 2012 Never satisfied, I decided the indirect lighting from the previous attempt was underpowered, leaving my humans in more shadow than I desired. I assumed this was the fix to my evening look problem. I also swapped out the flat water slab in place of a dynamic terrain surface. Even at 4096 resolution it looks choppy due to the scale of the total project. Distinct foam is added along the coasts but at a severe render time hit as it is a duplicated terrain with transparency mapping to keep the foam along the edges only. SLOWWWW especially with clouds. Remember that the scene is lit with EGDLS so there are multiple radials floating all around interacting with that transparent foam terrain!!! For my tastes this was about the limit on what I could make Bryce do using only Bryce elements. To go any further I'd need all new vegetation. Leaves are still cardboard.
4. September / 2012 By this point in time I was becoming fed up with Bryce native trees. For one, when instanced, Bryce native trees slow down the interface much more than imported meshes. This is an odd but true fact. So I decided to build my own palm trees and other veggies. This involves Carrara, Bryce, TrueSpace, and UV Mapper. I found I had much more overhead, I was able to paint much more vegetation without running out of memory which is why there are no visible vacancies in the vegetation along the ground. This is because I began to play with the concept of Eco-Tiles. Tiles are single meshes comprised of many small objects glued together and uv mapped so that they retain their textures but behave as a single model in Bryce so that the instance lab can rotate them (IL doesnt rotate grouped objects properly, only single meshes due to a bug). The result is more consistency.... No holes. But in a few weeks I began to feel these palm trees looked more like tall ferns, not really palm trees. Still illuminated with EGDLS
5. November / 2012 Unsatisfied with the thin leaves on my previous palms, I decided to rebuild them but with slightly broader leaves. I also decided to skip the foam and concentrate more on caustics at the water's edge. Once again a happy couple enjoying the heat. Leaves still look a bit of cardboard. There is still too much shading under the canopy which makes the image look of evening instead of broad daylight. Argghh I also decided after a couple of weeks that these new palms were no better than the previous ones. I decided to rebuild my veggies from scratch once again. Lit with EGDLS
6. May / 2013 Here we go again. Still lit with EGDLS. I started for the first time with new terrain design. The trees have been redone, this time I like them. The ground level veggies are redone as well and comprised of a few species of plants and grass. Though it provides ground cover, it doesn't vary enough in species to look natural. There is also the issue of lack of translucency making the leaves look of cardboard, especially for those plants closest to the ground. I am "almost" happy with this version. but you know me, I always want MORE MORE MORE.
7. October / 2013 Another new terrain base model. The trees I kept from the last build but the ground level veggies are completely new. There is a much greater variation of species here, and there are fewer trees in the canopy allowing more light to reach the ground level plants. The leaves have all been made ruffled in some way, and the textures are higher resolution. The more important change is that the leaves no longer look of cardboard, thanks to ambience. Typically, ambience ruins renders due to its perfect uniformity. Yes we know that leaves are transparent and should have a slight bit of what appears as glow to them. How I did it was to use a gradient image map for each leaf type, that gives me ambience glow on the outer edges of the leaf that falls off toward black near the center of the leaf. Because the EGDLS is doing all the hard work, the ambience is used as a tweak and it works well. The result looks great to me. Feedback is as always appreciated.
cont...
Wow Rashad, I don't know much about your project but your renders are awesome. Good luck.
@Rashad, what a wonderful set of renders. Indeed, some of the earlier ones have a strange large grey shadow area. I think the last two are both very good. I think I like the scene (with the sea in the background) in the earlier of the two slightly better than the last one. But the foliage is wonderful in the last one. I do however think that the happy couple is a bit white and those in the earlier version have a colour that better fits the scene. Perhaps you can modify this by shining an additional, warm coloured light on the couple (and have it only affect the couple)?
Also, a small creepy crawler on one of the big leaves on the front might make the scene even more realistic.
Nevertheless, I stand in awe for your perseverance and skills! You must have spent many, many hours, slowly improving and improving the series of renders. I would have lost my patience ages ago!
I hope to upload a new scene with my 'clothing grass' in the future (though probably not really 'near').
Next in the series 'use hair ( or in this case clothing) as grass. I made this image with my clothing-to-grass method, using instanced grass in the foreground and a terraintexture in a similar colour in the background as per the advice of Horo. I tried to do the trick that Rashad explained, but when I first did it with a grass with no texture, emptiness was the result and when I tried it with textured grass, my machine took way to much time to export the multiple grasses as one object and I did not have the patience to wait several hours.
This view was also quite a challenge for my poor laptop. After an attempt with only the horse, the Daz to Bryce bridge decided that it didn't like me and stopped cooperating. So I had to export horse and lady as object and import in Bryce. This only worked after deleting several copies of grass and a number of trees. Then still, Bryce went 'out of memory' and not all textures on the clothing of the lady were present. However, I could make simple colours in Bryce, so that is not a problem.
The horse is Millenium horse by DAZ, the lady is a variant of Genesis with some dial twitching by me. Clothes are a set that I got with Genesis (I think Magus jacket, shorts and shoes). Texture of Millenium horse was also included. I used a premade 'leaning' pose, but modified the arm a little and used a pre-made 'grazing' pose for the horse.
I am not completely happy yet. I have to get to grips with Rashad's method, or I'll need to render foreground and background separately, because I would like to add some flowers or bushes and some instances of leaves that have fallen to the ground. I am also not able to save this file, Bryce keeps crashing when I try.
Nevertheless, I am generally happy with the way this worked out.
I would really like to hear your feedback.
I also have a question: does it help to kill as many processes as possible on the computer to give Bryce more memory? If so, I will kill things such as Google Chrome and Explorer that I tend to have open and make a lot more memory available.
Almost forgot to mention a number of things:
* rendered in Bryce 7.1 Pro with premium settings, IBL using the sky as lightsource, True ambience, TA scattering correction and boost light.
* size of document without lady and horse: 191.304 kB
* Render time around 6 hours (not bad, I think) with priority 'high'
If you want to know more, I'll try to look it up!
@hansemar - I think you did a great job here. In Bryce, the sky is not the limit - but the memory is. Open the Task Manager and under the Performance tab you find the memory usage. This will show you if terminating certain processes help preserving memory. The document size is no indicator how much memory the Bryce scene uses because the file is compressed.
I don't know what your intentions were. It looks as there is no sun in the sky, there are almost no shadows. If you don't use TA, you will get strong sun shadows - you might use soft shadows and render premium if you are prepared to wait for the render to conclude. And since you ask - the Bryce default sky is a bit on the pink side. Skies are usually blue - or greyish-blue. Look in the sky library, there are better ones. Try Lazy Afternoon. It is in the Installed library under Daytime the first one: top row left. Nevertheless, I repeat, the scene looks already great as it is.
Hansmar,
Thanks for the feedback on the Volcanic Archipelago project. I think the dark gray "shadow" you are seeing is supposed to represent the wet sand along the water's edge. The alpha mapping makes the reflection subtractive from the diffuse color. I now use a more subtle effect that's less drastic.
On your most recent upload I agree with Horo. Right now there is plenty of pleasing indirect light but there is no key light. What we need perhaps is the default sun bright enough to provide the key lighting and shadows.
On making grass. I wrote out the tutorial earlier but I did not provide pictures which are probably necessary. From what I can see you going for you will indeed benefit from what I am proposing. I have written the tutorial this time with visual aids. Let me know if anything is yet unclear.
As follows:
1. Here are a series of grass blades I created in Truespace. The blades are already UV mapped. All six blades share the exact same UV map. Make sure that in the attributes that each blade is set as neutral or positive, not negative or intersect. The material applied is default gray and must remain that way for many good reasons I will explain below.
2. I select the ground plane to paint upon. I go into the IL (Instance Lab) and I then disable Instancing by clicking on the icon indicated in the image below. I don't want to paint instances, I want to paint real geometry. Trying to export a group that has instances in it will CRASH Bryce. More on that in a second.
3. I then go into the Brush Editor and I assign the six blades as sources. I also ensure that I have applied rotation and scaling variance for each blade. Most times set these parameters when loading the first source and the subsequent sources will also have the same scale and rotation settings. Click onto the pie chart to select which source you want to edit.
4. Now back to the Painter Screen. Paint a small clump of grass, then exit the IL.
5. Now you have the group Unnamed. You need to Un-group it, then re-group it. It's new name is now Group 1.
6. Export Group 1 as an .OBJ. Because the blades all share the same default gray material the export is lightning fast.
7. Open Free UV Mapper. Load Group 1. Ignore the notification about an error loading materials, it is irrelevant. Ensure that the blades UV mapping hasn't gone screwy which I doubt that it would. You will notice that the UV map is upside down, because there is a bug with Bryce where it flips UV Maps vertically when it is exporting.
8. Export Group 1 using the settings indicated below. Keep the same name Group 1. What you want to make certain of is that the option to export materials is disabled. If your UV map has been flipped by Bryce during the export then you might want to enable Vertical flipping as you export from UV Mapper so that it aligns properly once back in Bryce.
9. Now back to Bryce. Import the Group 1 you just sent from UV Mapper. When loading an OBJ, Bryce gives certain options. You want to use the Materials only option.
10. Now the magic happens. Bryce needs a second to import the mesh. But you will notice that indeed the multiple blades has re-imported as a single mesh. Name this mesh Clump 1.
11.. Now we rinse and repeat. Go back into the IL, but this time instead of loading the separate blades as the source, you will instead load the Clump 1. You will paint Clump 1 onto an area then you will exit the IL. Remember that we already disabled instancing painting so you are painting copies of real geometry.
12. Just like before, you are presented with a group called UnNamed. Export it to UV Mapper. Then export it out of UV Mapper again with materials disabled and with Vertical flipping enabled.
In this way you should be able to create tiles of grass as large or as small as you'd like.
13. Here is an example of my grass tile's relative scaling compared to an imported M5. Using my finalized grass mesh, I can paint grass as far as the eye can see in all directions at once without running our of memory. Tiling is tricky because if the terrain has a lot of bumpiness in it then your tiles will need to be much smaller otherwise you will see the edges of the tiles floating above the ground. For Volcanic Archipelago The terrain beneath the veggies is almost perfectly flat so my tiles rest along the ground from center to edge, Had the terrain been more dynamic, I would have needed much smaller tiles. Is this making sense? You can paint instances along vertex normals by pressing ALT while painting. You can then alter the Navigation tool to Object, and then you can sink the grass clumps into the earth so that nothing floats above.
So in my case I start with a single blade, I then create a couple of clumps and I export those clumps as objects. I then re-import those small clumps and repaint them to create larger clumps. Then, when ready, I export the clumps as a super clump. Once I get a clump pf a reasonable size, I'm good to go.
Almost forgot. You will notice that the grass blades I am using have extremely long stalks. This is because when painting tiles over a dynamic terrain some parts of the grass tile will touch the ground while other parts will appear to float above the ground. I have found that one might need to slightly submerge the tiles into the soil, and for that to happen one needs to have the length from the grass blades.
MEMORY ISSUES TO KEEP IN MIND:
There is a bug, a nasty one. After a scene has been saved and reopened memory usage will soar. As Horo suggests always keep the Task Manager open while working. Watch the way memory usage increases each time you paint more instances. Know that if you paint what appears to be 100mb of instances in the original session, once closed and re-opened the memory usage jumps to about 500mb. There is a roughly 500% mark-up for instances. The reason is because the link between materials is broken after saving and reloading. For example, during the original session if you alter the materials of the source the child instances would update automatically with no cost on memory. But after saving and reloading, you will find that altering the materials of the source does not affect the child instances as it should. Instead one must manually alter the materials for the instances. This results in much more memory overhead and is the equivalent of a memory leak. Its as if the whole idea of instancing becomes partially broken, the instances become partially real at least in terms of the way the material lab views them. Before you save again you must make certain that the source and their instances share the exact same material settings. Though the materials link is broken, if you don't ensure they share the same material before saving the internal logic of the scene will fail and the save will fail.
One last tip. The Undo Buffer holds the last 15 actions. Large actions, such as painting instances use up more memory than small actions. The only way to truly clear the buffer is to close then re-open Bryce. But that isn't always what we want. You can reduce memory occupied by the buffer by replacing memory intensive actions for those which are easier. I tend to create a sphere, I then resize it about 15 times, then I delete it. I find that on the 15 re-size the memory usage for the scene will drop dramatically, often by hundreds of mb.
cont..
Thanks a ton! I do enjoy working on it. It has taught me a lot.
@Rashad - would it make sense to put it together? I think it's a great tutorial.
@rashad,
Thanks for the very extensive explanation. I'll have to give it a try again. First read it two or three times, to see whether I understand it ;)
And I agree with Horo: a nice video tutorial would be great!
@rashad and @horo,
I agree with your remarks about the lights. I actually did not give that too much thought. Next try should include some more directional light indeed.
Hope to produce a new version later!
Horo,
Thanks a ton. Yes, it would be good to write it out fully as a real tutorial. I have a couple more reference images I want to post now to more or less complete the tutorial. Thanks.
Hansmar,
I love your vegetation from fabric. It looks like it yields the capacity to produce lots of ground level plants of several types. I say keep going with this there's no limit to what you might accomplish. Maybe we can work together, you send me a source grass and I send some to you so we can play around with the different methods. I will send you a pm with my email address if you are open to it.
Back to the tutorial.....
14. When we left off last I had just re-imported a small clump from UV Mapper. I will now use this as a new source, and I will create my first real tile. The grass you see in this current shot is a single mesh constructed as explained above with a few back and forths between Bryce and UV Mapper. I wanted a grass that was patchy and not 100% consistent. The good thing about this is that because the grass is a single mesh object it will rotate properly by the IL ( the IL cannot truly rotate grouped objects) Notice how much width the tile has compared to the scale of an M5. It costs around 400,000 polygons.
15. I have now loaded that tile from the previous shot as my new source. Now I have gone back to the IL and re-enabled instancing since now I am ready to cover some ground. I also want to keep the 360 degree rotation, but I want to remove all scaling variation because I don't want grass at all different scales all over the shot. As you can see from the shot, I have no problem covering huge areas with grass. Keep in mind that there is just as much grass to the sides and behind the camera as there is grass in front of the camera!!! This is a lot of grass and I could go on much further with memory to spare.
16. For some perspective, I have also included one of the "Tiles" used for the Volcanic Archipelago project. I decided to extend the idea to include not just grass at the ground level, but flowers and smaller trees as well. Notice the M5 at the center to provide a sense of scaling.
17. Here is a quick reference to the final UV Map for the Tiles. Notice how I have carefully assigned each leaf and stem its own area of the image. The sample uploaded here is at a sample resolution, but it's enough that you can see that it relates to the tile in the scene.
18. The tiles I am using are wide, so they cover lots of ground. Here is another quick image of VA to show how these tiles once construction can be painted over a terrain and appear natural similar to ecosystems. The approach of tiles isn't perfect and runs into lots of problems on terrains that have lots of dynamic bumps. But more on that at some future time.
Thanks for reading all this. It is about 6 pages total of text so it's no surprise it is challenging to get it right. It took me a couple of years to figure this out on my own.
Almost forgot to remind you that SHOW AS BOX is your friend!!!!!! Use this to reduce the wireframe detail and it will keeps things moving faster. There is a cost to working with high poly objects. Bryce reactions can get slow when editing an Unnamed that has millions of polygons, but once you have finished editing large groups the scene navigation speeds back up to normal. but only when Show as Box is enabled for all large groups.
Cool, just tell here once the tute is done. I'll gladly peedee-ef it.
@Rashad, Thanks. Indeed I have been thinking of other options, like stems and leaves. Advantage of the method (once you get the physical reaction of the material under control) is that you can make a starting 'cloth object' and modify it multiple times by letting MD run shorter or longer periods, slightly modifying the material, adding some wind, etc. In that way you easily have a variety of otherwise similar objects.
I'm really interested in working together, so, send me your e-mail and I will send some of the objects I made. Most have some groundfloor that can be deleted before working further with it, but that can also be buried beneath the ground.
You're examples look very good to me. I tried your method with my objects and got it working somehow. However, still struggling to get a reasonable spread of grass on the ground. I think I made it much too dense. Will redo it later, after I have finalised the present version.
Another issue with your method and my objects is the fact that UV Mapper can easily be out of its depth with too many leaves. But, practice makes perfect and I will keep perfecting.
For example: now my groundfloor is laying on a surface. It could make it hanging in the air (from some support) and then put the blades underneath, in that way ensure easy straight downward hanging 'grass' that can be modified by wind.
I am also thinking of much wider and not so straight 'leaves', etc. etc.
Anyway, feel free to play with objects I will be sending you. And I am really interested in playing with your 'tiles'!
Hans
Rashad's tutorial above is now assembled to a PDF document and I will put up the link once Rashad had the time to peer-review it.
Hi all,
I kept working on a similar scene to my last example.
@Horo suggested to use grass blades only in the front part of the scene. I followed his advice.
@Rashad Carter had a great tutorial in parts for making clumps and then instancing those to avoid 'overloading' the instancing option. I practiced and tested and indeed, it works!
@Horo noted in my last submission the lack of sunlight. He was right.
I myself noted that the grass might have been to big and too dense (which you don't see, but influences the memory a lot, I think).
@Horo and others also indicated I should use Large Address Aware.
So, I used the same terrain and started all over.
Used already existing clumps of 'grass' from Marvelous Designer to build one bigger clump, according to Rashad's advice. Then started instancing, ensuring a lighter spread. Also added (and instanced) some trees. Added a lion (Mil Big Cat from DAZ 3D) and a women (Genesis version 1, DAZ 3D) and a horse (Milhorse, again from DAZ 3D). Also some bushes (of which one is s tree) as well as several 2d faces for some old autumnleaves in the grass (instanced).
I now also added some sunlight to ensure that ther is some shadow.
So, the result is 369984783 polygons big, but still renders quite quickly at Premium settings (Sky as IBL, True Ambience, 256 rays per pixel). Rendering done in less than a day.
I was not completely happy with the brightness of the result, so, after exporting to HDR, played around with Picturenaut to tone map and modify the result somewhat. I like the present version.
More improvements are certainly possible, e.g. by using the technique described by Rashad to use 'a gradient image map for each leaf type'. Texturing, other than playing with the procedural textures, is really not my strength yet. There still is a lot to learn!
No issues with lack of memory anymore, but also no occasion where Process Monitor showed more than 1.6 GB memory need for this file, so apparently LAA was not needed.
So, again, I am happy with all feedback and advice. I will start playing with Marvelous Designer in an attempt to make other kinds of plants (that can be varied quickly). No idea where this will lead!
@hansmar - huge improvement. Good idea to export the render as 48 or 96-bit and use the tone-mappers of Picturenaut. By the way, Picturenaut accepts Banterle's plug-ins for HDRShop as well. Gives you a couple more operators.