Attached Link: http://www.cgtalk.com/showthread.php?p=2197421#post2197421

This is the original tutorial
http://freespace.virgin.net/david.markwick/VueTut1.html

richcz3

Message edited on: 04/22/2005 16:26

Thanks, still struggling with this stuff. The saying, can't teach an old dog new tricks is proving true ~L~ But thanks for all the help.

Yep multiple image maps is the way to go for ultimate control, indeed the only way if you want different spreads of different objects. The tutorial was only ever a springboard to get people thinking about the possibilities. One tree & one image was the by-word there :D

The above is a test picture...

Message edited on: 04/24/2005 14:11

Hstewarth, is your material mapping set as Object-Parametric? The world mappings don't seem to work out for this kinda stuff.

Yes it is.. I change it to several different variants. Basically so far what I found is that if my underlying material is a mix material and then I change the ecosystem scaling, the problems happens. I found it best to the following steps.. 1. Apply distribute map as color map to terrain 2. Change to ecosystem and using distribution map as map to distribute the ecosystem. Test and every thing is ok here 3. Change the underlying material to a mix material that is distributed by my distribution map. Test and every thing is ok here 4. Adjust the scaling of ecosystem objects Test and problem occurs.. The distribution map is simply a white square with a line down the middle for the road

hstewarth Is that tree showing up in a dark gray or black region of your image map? If it is a light gray part you may want to consider making the colr value on that strip darker. richcz3

I think richcz3 is right. I just tried following the 4 steps above using a black and white dist map, and everything went perfectly... even the scaling in step 4. Changed my map to use shades of grey, and the plants popped out along pathways.

If plants pop out on shades of grey that they're not supposed to, its a matter of fine tuning the filter curve.

Actually the tree is in "Pitch Black" part of the image map.

There is absolutely no grays areas in the map. When I get home, I can update the thread with image of actually map. It simple - drawer a small black line down the middle of 1024x1024 white square.

step #3 is very important in my test.

Everything works if I used the distribution map as underlying texture for the terrain. Only when I change it to use mix materials with a distribution map under it.

I have work around the isuse by always assuring that the ecosystems are done first but I should not have to do so. Is this possible, if my black is not 100% pitch black in rgb colors ( no where near gray ), could ecosystem be that sensitive. Does it have to be RGB(0,0,0)

Message edited on: 04/25/2005 13:30

Try changing the associated filter to a step filter in that case, with the step happening about half way across the filter.

If you have the default filter, or even a curve filter, very slight variations from absolute black could generate a random tree. If you have a step filter that will only trigger tree generation at mid-grey, you should be OK.

Message edited on: 04/25/2005 13:39

I played around with the filters and made them similar to ones in tutorial.. it had multiple steps and most of very low and then steps up steply. Maybe I am doing the image map distribution wrong on the materials. One thing I did notice if I load the distribution map as texture map, it not fit the terrain correctly like when I orginally added it to terrain. I had more than one line on it - but this was only after change it to mix materials and then changing the ecosystem.

Sounds like you might be running into a problem with map rotation. I started out by using my black/white map to help shape the terrain (wanted the walks to be flat, ground to be lumpy). That worked fine. Then I used the same image to drive the distribution of 2 materials. The distribution didn't match up until I rotated the driving map 180 degrees. When I added my map driven ecosystems, again I had to rotate the distribution map.

I believe I attempted to rotated the maps but not sure if I rotated both.. I will check them out tonight.

If you're using a pure black & white image (i.e. no shades of grey) try changing that filter next to the Variable Density material to a step filter. It looks as though very dark areas will still get the odd tree from here.

So far, it's not so bad for me, tho. Maybe that's because I'm only working with small stones and brush, while you're using big trees. Some spill-over is okay in a wilderness, but not in a park or garden setting.

DMM, I'm still lost when it comes to filters. What is a step filter (I can't find any that go by that name)? Also, how do you know what settings in the filters do what? For ex, "inverted clip under 40%". What does that mean? 40% of what? Guess I just don't get along with graphs.

BTW... I'm very happy to finally be controlling terrains, adding features, etc. Thanks to everyone for all the informative posts lately! Cool stuff!

I try a lot of stuff and still no go.. I am curious what exactly do you mean by step filter.. I try the one that looks like steps.. maybe I am missing something. any filter does not seem to work.

DigReal, I am desiring to work on a more complicated scene.. but I narrow down some problems to what I am finding now.. oddly enough I had better luck last night for some reason. One possibility, I accidently tilted my terrain and then do an undo on it. Is it possible that cause the problem. I could upload the scene to my FTP, but I planned to watch 24 recorded on my Tivo right now. DMM, thanks for you help.. Stewart

Right, without firing up my other PC & grabbing screenshots, a step filter is a filter that has zero value along the bottom till halfway along, then will suddenly jump up to full value (one) in one step, so that half the filter is grey and the second half is white. Imagine a grey scale along the bottom of the filter, going from black on the left through all the shades of grey to white at the far right hand side. These are your grey shades in your terrain image. Now imagine, on the other axis (the "y" axis) another scale, except that this one is "amount of trees". Zero value (bottom edge of filter) represents no trees, while the top edge of the filter (maximum value) represents lots of trees. The step filter will generate no trees for all image values from black to mid-grey, after that any grey tone up to white will generate trees. I will put up some pictures explaining more if anyone really wants them.

DMM, you're explanation of the filter graph was perfect. I think I understand completely now. Many thanks for telling it in a simple, straightforward way! After adjusting my step a few times, got perfect results. In fact, now that I understand, I'm looking forward to feathering at the edges. Thanks again. Stewart, I'm at a loss for any more ideas on what's going wrong at your end. I'm using tga for my maps (old habit), but don't think jpg would cause the problem. If you want to upload your scene, I'll try it on my machine and see what there is to see.

I can place the scene on my ftp site when I get home.. it will be about 6.5meg.. I will also try the tga format. But I also try BMP and it did not work. One thing I thought about is using a distribution map that is slightly wider and maybe it would resolved the issue. Sort of a kludge.. I am lost for ideas on it too.. Unless others are having problems like what I am having, I must be doing something wrong.. I just like to know what so that I can fix the problem.

Yeah, I really doubt it's the image format. I have dsl, so no problem downloading 6.5 mb. I'll watch for your file at the end of the day.

The file is actually on 3.2mb.. I have uploaded to the ftp site. You can get at the following address.. replaces * with t. I used a co-workers site and I desired to be respectable. Once figured out, I will remove the file and message. h**p://home.puffinsoft.com/hstewarth/Vue/distproblem.vue

Just want to let you know I got the file. Opened and repopulated the trees... same result you have. I won't have a chance to experiment til morning. Get back to you then.

Thanks, I am glad that you experience the same problem. Important thing out to find out if its me or V5I.

Good news and bad news! Found the problem... not sure of the proper solution. The problem is the size of your ecosystem objects in comparison to the resolution of your terrain. My guess is that there's some magical ratio we should follow for this kind of stuff, but don't know what that ratio should be. Basically, if we want to use maps for distribution, we need to give Vue more resolution to work with. I usually start out with 1024, which is why I didn't have the problem. Anyways, now you know where the problem is, so you can focus your attention on that. Have fun, Stewart!

DigReal, Thanks, I will do more experimenting with it tonight. So I assume that you talking about the bitmap of distribution map. it should be 1024x1024. That is interesting because I thought I used an higher resolution. did you have any problem if you change the texture to blend 2 textures using same distribution map. That was what was orginally causing me problems. By the way, I mention this problem on e-ons forum. Stewart

Not the bitmap. The terrain needs to be larger in the terrain editor. Click that little /2 a couple times to get it up to 1024. Of course, this mains a large file size than you were expecting... unless you change it to a procedural terrain (which renders slower). I didn't change your terrain texture to a mix, since I'm sure that's not the problem. In my example render above, I had 3 different textures blended and controlled by maps without a problem. But it was a 1024 terrain using 512 dist maps. Btw, using higher res dist maps has a different benefit. The higher the res of the dist map, the cleaner the seperation of the terrain materials. If there aren't any closeups, you can probably get away with a very small map.

Yep, a higher resolution terrain mixing image will yield best results, although there is a blending option for the mixed materials which essentially fades between materials instead of a sudden change. Sometimes this is not ideal, but more often than not moving the blend slider up to 100% hides a lot of blockyness.

Message edited on: 04/27/2005 12:41

DigReal, Also interesting, I think I try a larger terrain size ( I know at one time it was even 1024 ) I still believe e-on has a problem in Infinite, maybe the new release will help and hope developers at e-on will take a look at message. Wouldn't you agree, my scene should have work? But atleast I know ( thanks to you ) how to work around the problem. Stewart

With the proper filter adjustment for the gradient in the image, the rocks come close but do not cross onto the walkway.

richcz3

Message edited on: 04/27/2005 13:16

Hi richcz3, Yeah, just last night I tried playing with the gausian blur along with the blend options DMM just mentioned. Both produce nice results for situations like this. I was just as pleased with the results as you are (this is the kind of stuff I love learning about). Then the filter mixes even more. For something like a nature trail, it's even ok if the stones and smaller weeds move into edge of the path. That's the way it would be in reality. Stewart, I agree, from looking at your file I wouldn't expect a problem. Just don't know if its a bug or limitation of the math formulas. Either way, I'm really glad to know about it now, instead of in the middle of some big project with a tight deadline.

I have a question, does the gausian blur help out with distribution map ( I completely understand about its benifits on material map ). If the ecosystem logic is work, I would think you want a more exact map unless you want less frequent occurances in some of areas. Part of differences between my images and both of your images is that I am looking for a more close up image of the trail/road instead of an overhead image. Does this play into this.? I would assume I need larger textures..

Bad news... The stupid thing is still doing it, I made the terrain 1024 by the terrain editor as stated. Then press clear and populated the image and still puts the trees in to black area. I had enough with this, e-on is getting a bug report. DigReal, are you sure you are not running the beta.. because it still doesn't work. Dig you change the overall density or scalling of my scene. I have seen some cases it is ok at different density, the main issues is that it does not work as I enter and I believe its correct. Stewart

hstewarth
Yes the gausian blur is a key part for the distribution map. With the right filter settings there are no objects crossing. For that tree, you may need a greater area of blur or dark gray to remove it. Your right it does matter what scale you are using. I don't think it's a bug. You just have to plan the distribution map to manage that scale of tree better. Wish you luck

richcz3

Message edited on: 04/27/2005 19:57

I'm sorry, Stewart. I'm not very good at explaining what I mean (which is why I stopped writing tutorials).

This is actually more involved than my previous posts would suggest. Guess I'm just used to working with scale, so I overlook stuff along those lines.

First, I should've pointed out that when increasing the res of the terrain in the editor, you have to go to the size thing on the upper right and adjust it to match the editor. That is, if the editor is set to 1024, the actual size also needs to be 1024. And while I'm thinking of it, the inverse will also work. That is, leave the terrain at 256 and reduce the overall scaling of your ecosystem.

What I was trying to get at before, is that this isn't about size specifically, its about the ratios between your terrain, the stuff in the ecosystem, and the details on the driving map. In other words, the scene you set up has a ratio that appears to have too low a threshold for Vue to handle. That's why I suspect that instead of being a bug, this is a limitation of the code being used. Whether it's a bug or not, I DO agree that it's something e-on needs to work on. The scene you have isn't unreasonable... it should be able to work as is!

Anyways, until e-on does solve the problem, we need to work around it. Unlike graphs and nodes, this is something I'm used to dealing with. If you have trouble thinking in terms of 3d ratios, I'll be glad to try and help. As I noted, I'm bad at explaining... but I'm always willing to try to help.

Oh yeah, in case you haven't noticed... I'm absolutely positive this problem is about ratios. Not density, nor bitmap res, only ratio. Of course, that doesn't mean we can't use those as a work around.

I just check it out and I also came up conclusion that this is problem is related to scale. I took my same project and reduce the scale of tree to .900 and it appears to work. I also increase the terrain box side in window and when I this the trees got smaller and did not have the problem. I then increase the scale of the trees and once they got to a certain value, I had the problem. As a developer myself professionally, I believe this is a bug.. but it maybe limitation that we must work around. I believe with knowledge that I learn in last 2 or so days that I can work work around it. It would be nice to know the limits of what is involve. Good news.. I don't have to worry about that part.. and I can continue on with what I was planning... Stewart

Stewart, it looks like we cross-posted. But, yeah, what you're showing is what I tested, and made me realize there is a ratio problem. If you reduce the cherry tree in comparison to the terrain, it starts to work (or, if you increase the terrain in comparison to the cherry tree... same result). Anyways, it sounds like you understand about the ratios, so you know the work-around. Rickcz3, I like what you have going with only one big tree. Looks like everything is exactly what I'm hoping to do from this thread. I'm not sure yet if I understand what you're saying about where black meets white, since it's a gausian blur. Can you post a pic that shows the filter in use?

Looks like we cross-posted again. Hey, if you're a developer, then I'm happy to accept your definitions of bug verus limitation. :-) Either way, we have got to work around it, and I think we're making good progress. Ahd yep, in the last 2 days or so, I've learned a lot (from this thread alone), too. I'm really looking forward to turning all this into a real render! Thanks guys for helping to keep this subject alive.

Bug veus Limitation ( in my opinon ) Basically a limitation would be something that the software could not handle without significant designed changes. It may be weight that these changes would be worst than side effects of the issue. IE for example to fix this issue, the code would have to be change so that it would increase the rendering time signficantly. It may be something that was over look and one of e-ons developer will run it though a program called a debugger and find something wrong and change the code so that the problem doesn't happen. Not saying this is the problem: But one possibility is that there is calculation some where inside the calculations for ecosystem there is calculation that divides the size of object by size of surface and maybe that value gets too small and that could explain why its off. My next step that I was planning on doing after running into this problem, was to make sub sections with in other sections. I have recently learn that the best way to learn this stuff is to experiment and I finding that advice has help me out more than anything else. My initial post in this thread was base on isolated experiment based on the documention provided. Stewart Stewart

I agree with Stewart, this may be a case of trying to get the ecosystem to do something it wasn't designed to consider. My feel is that the purpose of ecosystems is to automate the scattering of a 'large' number of 'small' objects across a 'large' terrain. Where you are trying to place a small number of large objects then either manual positioning or scatter/replicate seems a better option. Mark

Mark, Actually you can still use an ecosystem for what I was trying to do.. just most of objects are outside of camera view.. It is all a matter of how you view it. The good news is that there is a work around on the problem. My big concern in my test, I was pretty much using all the defaults and it happens.. To a new user this could be an issue. Stewart

You also have to bear in mind that the Ecosystem is essentially replicating the same object, so for close up work a few individual trees will probably be better, with Ecosystem trees further away.

But if you using Vue Plants ( not XFrog ), when it replicating the objects, I believe ecosystem should make different variants of the trees The scene I currently working on, has a forests of trees done with ecosystem and also in the forground, I plan to include some XFrog trees close up. One thing I found out that was real nice last night. I created a terrain in terrain editor with desired river basin area. Exported it has image and in photoshop on seperate layer - black out the area of river basin, applied blur to it and use this image as my distribution map. I did have to flip the image 180 degrees to make it work correct. I am desiring to mix anonther ecosystem in the middle of this with its on distribution maps for the scene. I think the combination of ecosystems and image map distribution has a lot more potential.

Attached Link: http://img.photobucket.com/albums/v15/DMarkwick/FirstRender.jpg

I think the instancing algorithm is why Ecosystems render so fast.

Message edited on: 04/28/2005 09:29

My theory on ecosystems, is that if you notice in the plant editor there are parameters for change the look of a tree. I believe that ecosystem has a single instance of the object which is procedural object and for each variation of tree, it applys a slight different variation. Also rotates the object on the surface. If you read the manual, you notice that if you save an rock as vue object and include it ecosystem, it will also have different variations. If you decided to export an rock to another format, then you would likely only get scaling and rotation on the object. Very similar to XFrog or Poser objects in an ecosystem.

Hmm yep that sounds likely :) the rocks you mentioned may have colour parameters set by position as well.