Hi everybody,
I'm experiencing rendering-problems with an update I'm working on.
I've already posted this in my image-section, but since these problems are general and in maybe also in everybody's interest I thought I'd like to post it here as well.
Here's a crop of a testrender showing several problems, from left to right:
1) What's up with the intersection of the cloud with the populations?
2) Why do some flowers render completely and some don't?
3) Why doesn't the population rotate randomnly on it's Y-axis? All the trees have the same orientation :(
4) Why are the trees suddenly floating, while they were not before? I only added an extra PF-based terrain (3 in total now, was 2 before) with distance masking, all before the compute terrain.
This .tgd is 99% the same compared to the latest version and I don't understand these problems and haven't been able to resolve them.
Hope somebody can help.
Hope somebody can help.
Martin
maybe official explainations for those problems... that'll be cool
I must say, this is one of your finest compositions so far. Yes, even the advanced and knowledgeable users like yourself continue to have issues with the program. What is needed is a comprehensive TG2 manual with easy to understand explanations of the myriad of settings.
Well it would not be so bad if it worked right I do agree,
But then that is why they call em bata builds, sure would be nice to one day get to the point where a manual would be the right thing to have, instead of the next bug ridden update fix. But I know they are working on it, and I also agree this scene your working on is awesome bro. Sure would be nice to have a finished product that is stable though, humm, what a concept hahahahaha /winks.
Have a few tests I am going to do later today or tomorrow with the new build, but been to busy to play and spin dials yet for free and for fun. So TG2 has to have back seat :( 4 now.
DT
Are the trees populations or objects? I know you said the population doesn't rotate randomly but I just want to clarify that. Try changing the sit on terrain to the last node before the compute terrain and repopulate it. Just make sure you don't change the seeds.
Quote from: choronr on July 05, 2008, 04:06:54 PM
I must say, this is one of your finest compositions so far. Yes, even the advanced and knowledgeable users like yourself continue to have issues with the program. What is needed is a comprehensive TG2 manual with easy to understand explanations of the myriad of settings.
i agree totally :)
ha an antigravity tree!!!
could it be something to do with the displacements sometimes if the terrain is to displaced the populations will not map onto it properly ???
(it happened to me on the old beta)
Quote from: PG on July 05, 2008, 04:23:21 PM
Are the trees populations or objects? I know you said the population doesn't rotate randomly but I just want to clarify that. Try changing the sit on terrain to the last node before the compute terrain and repopulate it. Just make sure you don't change the seeds.
Thanks PG, but of course I've tried that...it doesn't make a difference :(
Quote from: lightning on July 05, 2008, 05:00:45 PM
Quote from: choronr on July 05, 2008, 04:06:54 PM
I must say, this is one of your finest compositions so far. Yes, even the advanced and knowledgeable users like yourself continue to have issues with the program. What is needed is a comprehensive TG2 manual with easy to understand explanations of the myriad of settings.
i agree totally :)
ha an antigravity tree!!!
could it be something to do with the displacements sometimes if the terrain is to displaced the populations will not map onto it properly ???
(it happened to me on the old beta)
The terrain consists of two fractal terrains (which are 2 powerfractals (PF's)). I added another PF to a total of 3 then, all before the compute terrain node.
And now the populations are floating, no matter which surface layer I assign to sit on.
It would be great if you could take and combine these terrains in Terragen 0.09; then, take them back into TG2 and continue your work.
TU, this is marvelous work. Sorry to see it so broken.
Have you moved your tree? Check that the object position is still 0,0,0.
I don't see anything clearly wrong with the trees in the fog layer, but it's hard to see what it's supposed to look like. It would be interesting to see those trees rendered from a closer viewpoint.
Could the similar object rotations simply be coincidence? 'Random' doesn't automatically mean 'different'. What if you change the population seed? (It will change positions too, unfortunately.)
We're doing what we can to figure out why some objects are not rendering properly.
Matt
Quote from: Matt on July 05, 2008, 10:36:18 PM
1) Have you moved your tree? Check that the object position is still 0,0,0.
2) I don't see anything clearly wrong with the trees in the fog layer, but it's hard to see what it's supposed to look like. It would be interesting to see those trees rendered from a closer viewpoint.
3) Could the similar object rotations simply be coincidence? 'Random' doesn't automatically mean 'different'. What if you change the population seed? (It will change positions too, unfortunately.)
4) We're doing what we can to figure out why some objects are not rendering properly.
Matt
Thanks Matt :)
1) I Haven't moved the object position. The translation of the object maker is set to 0,0,0 for all objects/populations in this scene. I have tried a negative altitude for the object maker, but it didn't make a difference.
Though the populated area is somewhat moved and increased, but very little.
The translation on the Y-axis in the populator is 0m which I think is correct.
However the actual altitude of the trees in the project is around -1600m. That shouldn't matter, isn't it? I've tried setting the translation to -1600m in the populator but that didn't make a difference.
2) I'll work on a more close-up render of the trees in the fog layer. In my opinion the shading transitions of the trees in the fog isn't smoothly and looks jagged (vertical lines).
But let's see if it appears in close-up.
3) I've checked 4 other seeds and the trees still have the same rotation. I've included an example in this post.
4) I've seen similar reports elsewhere on the forum, thanks for jumping into this matter :)
This tree non rotation seems to be an ongoing problem dating back to the TG2TP Free which I have ...I've had comments on a few renders I've done where they say,"Aside from the trees all facing the same direction....." and I know I have them rotating 360 around the Y axis.
Me too, Bobby.
Quote from: Tangled-Universe on July 06, 2008, 08:32:17 AM
1) I Haven't moved the object position. The translation of the object maker is set to 0,0,0 for all objects/populations in this scene. I have tried a negative altitude for the object maker, but it didn't make a difference.
Though the populated area is somewhat moved and increased, but very little.
The translation on the Y-axis in the populator is 0m which I think is correct.
However the actual altitude of the trees in the project is around -1600m. That shouldn't matter, isn't it? I've tried setting the translation to -1600m in the populator but that didn't make a difference.
The Y translation of the populator shouldn't make much difference (although very large values could potentially affect object spacing), so it's reassuring to see that it had no effect on your scene. However, changing the translation in the object maker (your OBJ/TGO Loader) should definitely change the placement of all the objects in the population, and it's a useful capability. I am surprised that there was no change when you used a negative value for this. I will test whether it's still working for me.
Quote
2) I'll work on a more close-up render of the trees in the fog layer. In my opinion the shading transitions of the trees in the fog isn't smoothly and looks jagged (vertical lines).
But let's see if it appears in close-up.
It does look a little unusual, I agree, so I look forward to seeing the close-up.
Quote
3) I've checked 4 other seeds and the trees still have the same rotation. I've included an example in this post.
In your cropped example the tree on the left bends in the opposite direction, so there must be a change in rotation. Unless this is a different model. The trees are also in different positions which I will assume is due to the change in seed. But I will also verify that rotation is working with populations of TGOs and OBJs.
Matt
okay, I may be speaking out my backside here, but is the levitating tree problem just effecting one population, of the type of pine tree pointed out, or does it effect all the tree populations. The reason I am asking as it is hard to see the bottom of the other trees of the same population due to a slight rise in the terrain between them and the camera, but a while back I used a model of a tree which had the origin set at the end of a root, causing the population for that tree to set above the terrain, with just the tip of the root touching the ground, the other populations of different tree models sat on the surface. Perhaps when the additional PF based terrain just lowered the intervening terrain enough to see the levitation.
Just a thought, slap me if I am way off.
SeerBlue
I'm looking forward to see your findings Matt.
I'll dive again and deeper into more translation settings to get the models to sit on the terrain correctly.
It seems that changing the seed of the clouds etc. resulted in better handling of the fog.
I see your point about the rotation, it indeed bends the other way, but if you compare branches and shapes they all look (pretty much) the same.
Quote from: SeerBlue on July 06, 2008, 12:57:56 PM
okay, I may be speaking out my backside here, but is the levitating tree problem just effecting one population, of the type of pine tree pointed out, or does it effect all the tree populations. The reason I am asking as it is hard to see the bottom of the other trees of the same population due to a slight rise in the terrain between them and the camera, but a while back I used a model of a tree which had the origin set at the end of a root, causing the population for that tree to set above the terrain, with just the tip of the root touching the ground, the other populations of different tree models sat on the surface. Perhaps when the additional PF based terrain just lowered the intervening terrain enough to see the levitation.
Just a thought, slap me if I am way off.
SeerBlue
I see what you mean and it makes sense. However, the origin of these models are above the trunk/roots.
Okay, I just thought I would throw that out, as far as random rotation, I am using 3 free, and it works fro me, do a test with an object that has a shape that is easily noticed when it is rotated, for example I made just a post with four arms at the top and placed a different shape at each end of the arms, loaded it in terragen and set random rotation 0 360 and it does it, I think the problem is most trees models are relatively balanced, so they don't appear to be rotating, I'll try to attach an image, but my connection is flaky today and keeps dropping,SeerBlue
Quote from: Tangled-Universe on July 06, 2008, 01:51:06 PM
I'm looking forward to see your findings Matt.
I'll dive again and deeper into more translation settings to get the models to sit on the terrain correctly.
It seems that changing the seed of the clouds etc. resulted in better handling of the fog.
I was looking at the fog again and it could simply be that the fog density varies more than you think, slinking between the trees and hiding some trees more than others.
Quote
I see your point about the rotation, it indeed bends the other way, but if you compare branches and shapes they all look (pretty much) the same.
If it bends the other way, that can only be explained by rotation or flipping of the model, but for Terragen to flip the model there would have to be a weird bug. The branches and shapes do have a lot of rotational symmetry, but that's just the way a lot of these Xfrog models are :) That symmetry is not entirely unnatural, either.
Matt
I've run some tests with your T-U_Finally_Spring_v3.tgd and I haven't found any problems, but I had to take some shortcuts. I didn't have your .tgos to work with, so I substituted one from the Xfrog Plants for Terragen 2 sampler. Changing the Y coordinate in the TGO Reader shifted the vertical position of the objects without any problems. Even if there is a problem with your .tgo, I don't see why the vertical position change doesn't work.
I couldn't get my plants to appear in the same places as yours, so I don't know if the seed has changed, or the object positioning problem is affecting that. But they were in the same approximate region of the image and they all sat on the terrain as they should.
Can you find out exactly which population is wrong, and then try 1) using a different object instead, and 2) using the problematic object in a population in a different scene?
Matt
Quote from: Matt on July 06, 2008, 03:35:53 PM
I've run some tests with your T-U_Finally_Spring_v3.tgd and I haven't found any problems, but I had to take some shortcuts. I didn't have your .tgos to work with, so I substituted one from the Xfrog Plants for Terragen 2 sampler. Changing the Y coordinate in the TGO Reader shifted the vertical position of the objects without any problems. Even if there is a problem with your .tgo, I don't see why the vertical position change doesn't work.
I couldn't get my plants to appear in the same places as yours, so I don't know if the seed has changed, or the object positioning problem is affecting that. But they were in the same approximate region of the image and they all sat on the terrain as they should.
Can you find out exactly which population is wrong, and then try 1) using a different object instead, and 2) using the problematic object in a population in a different scene?
Matt
Hi Matt,
I'm glad to report that the majority of the problems are solved :)
First I thought that lowering the object maker's altitude didn't work because the populator created more and also differently placed instances after lowering the altitude.
Somehow after a restart, then first populate and then adjust the 'offset' it seemed to work :)
It was the EU35m population of the foreground trees which gave the problems and I remember "fooling" TG2 to duplicate an other populator and set it to load the EU35m model.
It then worked for one session of TG2 and then the problems rose again.
This must sound pretty strange and unbelievable, I know.
I've also more or less solved the cloudproblem and I think it was just an incident. All the renders of different seeds gave better results.
The last two days I've spent on rendering the left part of the final image, so the last part should be finished by tomorrow or so... Can't wait to share the final :)
Martin
That's good news :) It may be that there are some bugs causing things not to update when they should. The populator used to update whenever anything even remotely connected to the population changed, to avoid that sort of problem. It's possible that some recent changes to make it more 'intelligent' are complicating matters. I suppose the most important thing to verify is that things render correctly after reloading your project (after you've made things right, of course). Does that seem to be OK?
I definitely look forward to seeing the final. This should have been Image of the Week already! ;)
Matt
Quote from: Matt on July 09, 2008, 11:18:40 AM
That's good news :) It may be that there are some bugs causing things not to update when they should. The populator used to update whenever anything even remotely connected to the population changed, to avoid that sort of problem. It's possible that some recent changes to make it more 'intelligent' are complicating matters. I suppose the most important thing to verify is that things render correctly after reloading your project (after you've made things right, of course). Does that seem to be OK?
I definitely look forward to seeing the final. This should have been Image of the Week already! ;)
Matt
I understand, that could be possible indeed. If I run into it again I'll report it as well.
Is it still correct that the objects reload after copy and pasting a node? It is half as annoying as before because TP4 loads and updates models much faster than the previous TP's :)
The final is nearly finished with rendering and I consider to make some extra slight improvements, but I have to consider it finished at some time. You're right about that ;)
Thanks!
Martin
Finished! ;D
http://forums.planetside.co.uk/index.php?topic=4231.msg46225#msg46225