Hi all
I'm still fairly new to T2 hence the silly question. I use Lightwave and use Lighting Probe images ( chrome ball ) to help with the reflections etc in my scenes. I wondered if I can create these HDR chrome ball objects in T2.
TO test I added a sphere, gave the sphere a reflective shader and this is what i get. I'm trying to achieve a very reflective ball. I'm using the free version of T2, maybe this is accurate in terms of reflections, but it seems dark.
Thanks
You can increase IOR from 1.33 to 100.
:) thanks. I thought IOR was for transparent objects its why I overlooked it! :-[
You're welcome, my pleasure :)
Cheers,
Martin
Martin ( same Martin who presented the webinar over the weekend? )
If so thanks again... I enjoyed it, its why I'm here battling away with a chrome ball :)
I'm getting a strange artefact on the render. I would expect the ground reflection to have a gradual change not the hard colour transition that i pointed out '1'. Also the reflections seem to be aliased '2'. I've attached the tad also... maybe i'm not setting it correctly or its a result of the free version limits?
Yes that would be me. Happy to hear you enjoyed it!
I have to think a bit about the first issue you mentioned.
That particular "band" looks funny.
I have a possible solution in my mind, but I'm not really sure and I can't remember the exact name of the setting anymore.
Both would probably only be confusing if I'd try explaining it.
About the second issue though, you could try the following:
1) you could start with setting reflection roughness to 0
2) the aliasing in the reflections could be two things. You can enable soft shadows in the sunlight node or improve the calculations of the reflections (see 3)
3) increase the "ray detail multiplier" inside your render node. This setting changes the detail in the micropolygons that are seen by the reflection rays. The multiplier multiplies the render detail setting from the renderer with the "ray detail multiplier" setting.
So, for example: if you render with detail 0.7 and "ray detail multiplier" @ 0.25 => 0.7 x 0.25 = 0.175.
In other words, the geometry/surface seen in the reflections is based on 4x lower quality than outside of the reflections. Simply put.
Therefore, the max number for the ray detail multiplier is 1.
(double click on the + symbol of the render node, then you should see the "render subdiv settings" node (grey) and inside that node increase the ray detail multiplier from 0.25 to 0.5, for starters)
You may understand now that whenever you render something with reflection that you 1) need sufficient render detail 2) enough ray detail multiplication, to get good results.
The Band is due to the GI prepass settings. Removing GI gets rid of it (simply select the Enviro light and de-activate it) but you may not want to do that. Try the same scene again but with some landscape and clutter in there, the effect may not be so visible.
I don't have a solution for the rough AA reflections.
cheers
Richard
Quote from: cyphyr on January 07, 2013, 08:00:30 AM
The Band is due to the GI prepass settings. Removing GI gets rid of it (simply select the Enviro light and de-activate it) but you may not want to do that. Try the same scene again but with some landscape and clutter in there, the effect may not be so visible.
I don't have a solution for the rough AA reflections.
cheers
Richard
I think the rough AA reflections are due to the combination of low render settings and the ray detail multiplier as I tried to explain.
I did know about the band being caused by GI, thanks for that :)
It would be a bit unexpected if prepass padding would work, as it's only a planar extension of the calculations and not spherical?
In other words: the calculations only extend to left/right om the image, but not backwards.
There's a setting for that in the render subdiv node, but it's only for shadows I believe and I also can't get to the name of it :/
Thanks for those tips... I got rid of the banding by turning off the Enviro tabs suggested. Also looked at the aliased edges by turning up the Quality Detail in the render node to 1 ( max in free version ) this seemed to help. Still a little choppy towards the edges of the sphere, but i guess increasing further would remove the artefacts.
Thanks I'll continue to tinker :)
Quote from: insignet on January 07, 2013, 08:45:41 AM
Thanks for those tips... I got rid of the banding by turning off the Enviro tabs suggested. Also looked at the aliased edges by turning up the Quality Detail in the render node to 1 ( max in free version ) this seemed to help. Still a little choppy towards the edges of the sphere, but i guess increasing further would remove the artefacts.
Thanks I'll continue to tinker :)
Ok good to know this worked.
How about the "ray detail multiplier" setting I told you about? Tried that?
Rendering @ detail 1 will create a tremendous amount of polygons which will be a lot smaller than a pixel and thus take much longer to render.
Bear in mind that your reflections still render @ detail 0.25 if you don't change the ray detail multiplier.
If you set the multiplier to 1 then the detail of your rendered reflections match 1:1 with the detail setting of the renderer.
A detail of 0.7 often suffices.
I couldn't find the ray detail multiplier. I looked in the render node but couldn't see it... :-[
Quote from: insignet on January 07, 2013, 09:35:42 AM
I couldn't find the ray detail multiplier. I looked in the render node but couldn't see it... :-[
In the node network click on the "renderers" tab/bookmark on the left.
The node network now focusses on 2 nodes, the "quick render" and "full render" nodes.
Suppose you're using the "full render" then double-click on the + symbol on the right side of the node.
You should see 4 grey nodes.
If you don't then move the grey nodes around, as sometimes they are right on top of each other.
Among the 4 nodes you should be able to find the "render subdiv settings" node.
Double click on it to open it and there you can find the "ray detail multiplier".
7 lines for 7 seconds of work, so don't be afraid of trying this :)
found it :) Yip it worked, smoothed out the aliased i set to 0.7. The only issue now is the shadow ( result of turning off the Enviro light to remove the banding) shadows are now solid black. I'll try to put a few other things in the scene to break up the banding.
I imagine that the setup you using at the moment is basically to see out how it works and iron out any bugs that may come up in your workflow. (Are you really trying to create a scene with a gray featureless floor and a blank sky? :) )
However it may still be worthwhile trying a more complex scene, something closer to your final desired output. The variation in texture, surface detail and cloudy sky may well hide many of the faults your seeing with the setup now.
Also if your aiming for a final output of an environment map then un-tick cast shadows on the sphere. The shadows cast by the sphere will be of no use to you.
cheers
Richard
Yeah... i tend to get confused when there's lots going on and loose focus on what i was trying to do. With T2 lots of numeric options / fields and for it to sink into my brain i have to go over it a few times :) I'm a simple fella
this is strange... I made a height field quickly and put the ball shape in there and it seems to reflect a lower res version of the terrain? You can see detail at the base then as it increases in altitude the goes lower res. Same results with Envir enable, disabled.
As Martin mentioned earlier this is (I think) due to the "Ray detail multiplier" in the internal network of your render node. Try higher settings in there but Martin is more qualified than me so I'd look to him for further advice.
Also you may want to search the forums for cube map creation (otherwise known as skyboxes). This avoids the use of the sphere altogether. Basically 6 images rendered looking forward, backward, left, right up and down stitched together seamlessly to form a cross (like a Christian one). Most 3d rendering software (including Lightwave) that supports environment mapping will be able to use a cube map.
hope this helps
Richard
Thanks will check it
Well boy Howdy!
I did not know there were internal nodes in the renderers. :o
It could not find them at first. They only appeared after I ran a first render and then I got a "+" sign at the right side of the node.
Much thanks from me too.
Update:
I went through this for a learning exercise(or exorcize ;D) I have the same banding. It seems to me to look like the detail in the refection drops off at the band. Perhaps because it is off camera maybe? Like a population crop or something?
Dammit... your orb looks better than mine ;D It's strange, the displacement detail gets reduced in the reflection, the banding and this issue must be connected in some way... but which way :-\
Ok, so *if* in this latest orb/sphere the "ray detail multiplier" has been increased to 1 and this didn't help then I can only think of one other cause at the moment:
The geometry which is being reflected correctly is all in frame or is only out of frame to the right/left of the camera.
The geometry which is being reflected incorrectly, at reduced detail, is *behind* the camera and probably can't be fixed using the options I suggested.
Quote from: Tangled-Universe on January 08, 2013, 03:21:41 AM
The geometry which is being reflected incorrectly, at reduced detail, is *behind* the camera and probably can't be fixed using the options I suggested.
if this is the case, move the camera very far from the sphere and change your lens to frame it up....
Obviously! slaps forehead. ::)
Of course, tried it and it works a charm ... only one possible issue. It would appear that holes are in parts of the terrain in front of the camera but facing away from it. A more detailed and populated scene may hide these shortcomings.
Thank you for your insight :)
Richard
ps the camera FOV for this is about 2deg.
What did you do/fix Richard? It's not very clear ;)
Just what goldfarb suggested, moved the camera about 70m away from a 1m sphere and zoomed in (fov at about 2deg).
In this way almost all of the terrain that is reflected by the sphere is in front of the camera but since the camera has a very high zoom factor (low fov) the foreground is not visable.
Easier to see from a tgd.
:)
Richard
I have to say I've always thought this was a very odd way of capturing data for environment mapping when you have a CG source. OTOH, I thought it was quite a clever way to capture real world data. Just had to make the comment :-).
Regards,
Jo
Thanks everyone for your help. ;) Like I said at the beginning I'm still learning, there's so many cool things on these Terragen planets, takes time to create new worlds :) btw on a slightly different note I found this interesting, Procedural generation talk for Elite Dangerous demo- https://www.youtube.com/watch?v=iTBvpd3_Vqk
Quote from: jo on January 08, 2013, 04:38:28 PM
I have to say I've always thought this was a very odd way of capturing data for environment mapping when you have a CG source. OTOH, I thought it was quite a clever way to capture real world data. Just had to make the comment :-).
Regards,
Jo
I agree. Somehow, intuitively, you'd tend to think that at the edges of such a spherical lightprobe you wouldn't have enough fidelity/resolution to make it suitable as a lightsource in other 3D packages.
There's a process where you take four chrome ball shots/render taken at 90 deg to each other and through some fancy photoshoping turn them into a single spherical projection environment.
scholarly complex article on the subject (http://www.scratchapixel.com/lessons/3d-advanced-lessons/reflection-mapping/converting-latitute-longitude-maps-and-mirror-balls/)
Slightly more sensible article (http://www.hello-napalm.com/tuts/hdri.html)
I agree though the simple cubic sky map version is so much easier.
Richard
Ha! All I wanted to do was make a pretty chrome ball just like the big kids on the playground. :)
So, am I right in thinking the fractal/procedural terrain just renders in view of the camera? And not off camera like, say, on the back(unseen) side of the mountain?
If that is the case, then it shows me how really elegant this software is.
Russ