Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Thank you Horo for the wonderful demonstration of the offset issue.
You're welcome, Adam. It was you, after all, who helped me in the old forums to make UE2 work.
I hasten to add that those 35° are with the default camera. I've read in another forum thread that it will not be so when using another camera. I will test this and also whether there is a vertical offset.
The offset is 35° but I don't (yet) know whether there is also an offset vertically. I'm working to get to grips with DS in general and UE2 in particular. Though for the Xmas 2012 give-away from the PA we (David and I) had an HDRI included for DS (and Carrara) and for that I had offset the HDRI by 35° to match the backdrop. But there must be a better way to correct for that.
The 'uberEnvironment2 IBL map axis is wrong' thread mentioned in post #1 goes into the offset values in some detail, and I think it's now fairly clear that using the UE2 IBL Transformer corrects them.
Anyway, here's the results of my (hopefully) final test:
- I created a 128x64 correctly prepared TIF and tried it in the UE2.
- I also tried it in the IBL Transformer.
- And just for fun I tried plugging a visually identical JPG into the UE2.
P.S. Horo - those Bryce examples are a very good demonstration of the colour mixing - now I understand why the test image with the UE2 IBL Transformer had tiny dots!
Edit - right/left labels on the images are the wrong way round!
Glad you got it working. I'm just skeptical about the 35° because I can't find any logical explanation to the number yet. I'd rather expect for PI or PI/2 rotation.
@3dcheapskate - great tests. Using the UE2 IBL Transformer is no option for me because I want to prepare HDRIs for a product we're working on.
I haven't noticed any change of time needed to specular convolve when the Phong exponent is changed. However, if you go for diffuse convolution, Banterele's plugin does the job much faster. Generally, a higher Phong exponent gives a less blurred result but the more the exponent is increased, the less difference you notice. At the time, we've implemented it into Bryce to go from 1 to 100.
I was doing some tests last night. I was completely shocked by the result and have yet to ponder over it. I have an idea what's probably going wrong, but it may be equally possible that I have got it wrong.
@Takeo.Kensei - yeah, it may well be that the 35° I measured are right for the very HDRI used and not necessarily for others.
I hesitate to say this, but would you believe Y-Rotate=-90 and X-Rotate=-90 ??
I know it sounds bizarre, but try it yourself.
I've repeated this test three times, rendering from different angles, because I thought I must be doing something wrong - I get the same results every time. But it really does need somebody else to double-check!
I did these renders through the 'Perspective View' using a sphere and cube primitive.
1st Image: Load the UE2 IBL Transformer and apply your desired test image - mine was the same texst image as I used in my previous post. The settings I used when loading UE2 IBL Transformer are in the last image.
2nd Image: Delete the UE2 IBL Transformer so you only have your sphere and cube. Now load the UE2 normally and manually apply the same test image with mode set to Ambient Only (I used the same procedure as in the OP, but with the correct image!). Then set UE2 X-Rotate and Y-Rotate to -90 and render.
I purposefully uploaded them as PNGs - download them, open the two images as layers in GIMP, and set top layer to 'Difference' - they're eaxctly the same!
3rd Image: was just a test using the values from this post.
There may be more than one thing that is a bit uncommon than just the offset. I used an HDRI with only one light source, 10° wide on a very dark - but not black - background. The dynamic range is 3312:1. One HDRI has the light at lat/long = 0°, the other one in the zenith.
(1) - Bryce render, at left light from front, right from zenith. Light and shadows match. The rib-cage is a model here, horizontal angle of view 200° (using the extreme wide angle lens). The ground is tilted by 11° to move the horizon down, otherwise the light source would be cut in two.
(2) - Carrara with the same HDRIs, no rib-cage, plane tilted as above. If the light comes in horizontal, there are no shadows. The sphere is lit on the other side. If the light source is in the zenith, there is a shadow underneath the sphere, which is correct.
(3) DS, camera focal length 16 mm. The rib-cage with the bright light is on the environment sphere. The image on the environment sphere has to be mirrored because it is mapped on the outside and the camera is on the inside of the sphere. I haven't done so here. Camera is set in the world centre and all rotation angles 0. I do not understand why we have the environment sphere offset by 17.5°.
The sphere at left is lit and there is a shadow that goes at right angles to the right, which is completely wrong. At right, the light shines down from the zenith and the sphere is correctly lit, less light on the lower part. The ground plane stays black for unknown reasons. I only swapped the HDRI.
I have specular convolved the HDRIs with a Phong exponent of 16 and repeated the test, there was no change even though the light got much wider.
(4) Finally, we have Octane using these two HDRIs to light a scene. The chess figures are correctly lit and shadows are cast away from the light. At right, the light is at the zenith and the shadow under the lying tower can be seen.
Perhaps I make the mistake of thinking UE2 is IBL. As it is working, it cannot be used to render with natural light. However, it can still be used to give nice ambient light as long as there is no prominent light source in the HDRI and the key light is provided by a DS light source.
How DS UE2 behaves reminds me of the flawed IBL in Bryce 6.0. To create the lights from the HDRI, the Monte Carlo algorithm was used and probably not correctly implemented. From Bryce 6.1 onward, the lights are created by the median-cut algorithm, which is more deterministic but takes a bit longer.
@ Cheapstake : OK I'll do my own test and post my findings here. The -90° is more to my liking because I already have some hypothesis on the source of the offset
For the third test you ran it may well be because some people did the same mistake you did
@Horo : UE is IBL, the algorythm used is just different from the others. Path tracing with Monte Carlo is always going to give more accurate results if enough sampless. Your results don't really surprise me. Renderman compliant renderers can get the same result but it depends on how it was implemented
Well one thing is certain from reading this thread UE2 is not right. LOL
Thank you for your reply. I can't say whether median-cut or Monte Carlo is better or where is the advantage of one over the other. As long as they are implemented correctly, both ought to give generally the same result. And then, there is still the chance that I'm doing something wrong, being the DS dummy I am.
Ok did a few tests and I'm still scratching my head
Don't want to give quick false conclusion but here what I found out so far
@Cheapstake : Your -90° on X and Y rotation are compared to what the IBL transformer is doing. So we just know at least that it rotates UE that way. Seems tp be right from my few tests but I'm not finished yet. And that doesn't explain the Environment Sphere being off
@Horo. I may have an explanation for your 17,5°. I'll rather say 18°
Just looked at the Environement Sphere UV and what I see there makes me think the angle is 180/10
To be continued..
Correct. I've simply assumed that, as tofusan says, the IBL Transforner is a fix to the well-known UE2 offset problems.
But also I think tofusan is correct, since the IBL Transformer results are what I would intuitively expect that the image I plugged in should give (which is simnply a left-right flipped version of a lat/long mapped environment test image) - each of the six coloured lights appears to be illuminating the scene from the correct direction.
Regarding the Environment Sphere, I never use it. I delete it immediately after loading UE2 (ever since Szark told me it was only a visual reference and didn't affect the lighting - edit: found the reference and added link). The Environment Sphere is irrelevant for me since I use my worldball for the environment.
Just looked at the Environement Sphere UV and what I see there makes me think the angle is 180/10
To be continued..
That looks very interesting, and I think I see the lead you're following - bad UV-mapping on the Environment sphere making it 1/20th (5%) off in azimuth ? Waiting to see your test results...
(P.S. This thread is where I started trying to use UE2, back in April)
Sorry! Post deleted - trying to take too many steps in one go.
(Corrected version (without the stupid mistakes!) of the test I posted here and then deleted is in post 51)
Could be, could be. I found 17.5° to fit rather nicely to my 35° offset. Wishful thinking, I reckon. I'm still handicapped by my ignorance of DS.
@3dcheapskate - interesting tests, again. Yes, the environmap is only for the backdrop since it is an LDRI. It serves well if there are reflecting objects in the scene.
Apologies on the previousdeleted post (if anybody saw it appear, change, and then disappear!)
I realised that before I start posting any more of my tests I need to be clear on my conventions (and even more clear that I'm actually doing valid tests!)
So here's the conventions I'm using:
DS4 has a little orientation cube at the top right of the preview window. Top is green, front is blue, left is red. Opposite faces are the same colours (bottom is green, back is blue, right is red.
For environment maps I always think in terms of the compass, with north dead-centre on the environment map, south, east, in there usual positions, and 'up' and 'down' as the other two.
When relating environment maps to DAZ Studio I imagine I'm inside that little orientation cube, so .
For my test image I made the north, up, and west colours match the front, top, and left colours of the DS orientation cube. But I made opposite faces complementary colours.
So in summary these are my conventions:
north (front) = blue
south (back) = yellow
east (right) = cyan
west (left) = red
up (top) = green
down (bottom) = magenta
First image is the DS preview cube, second is my environment map convention, and third is the various versions of my test image for different purposes. The light probe, reflection map and UE2 TIF (which is causing all the problems) are all flipped horizontally with respect to the first two.
Could be, could be. I found 17.5° to fit rather nicely to my 35° offset. Wishful thinking, I reckon. I'm still handicapped by my ignorance of DS.
@3dcheapskate - interesting tests, again. Yes, the environmap is only for the backdrop since it is an LDRI. It serves well if there are reflecting objects in the scene.
Hi Horo, Just noticed that you posted between my deleted post (43) and the subsequent post of mine (45), so I guess you saw the post I deleted, and you're refering to that? Those tests were invalid really, since I made a couple of mistakes and made too many assumptions without validating them! (I can reinstate the post if necessary - I still have the words and pictures)
Regarding LDRI - as far as I know it's fine plugging an LDRI (JPG) into a Poser diffuse light. I couldn't find a DAZ Studio equivalent, so I created my own DAZ Studio version of a diffuse IBL in ShaderBuilder, but it doesn't allow AO.
The reason I started playing with UE2 in the first place was to simply use it as an equivalent for a Poser Diffuse IBL, and as a short-cut to avoid trying to implement AO in my own light...
...and that's where my problems began... :bug:
Anyway, trying to correct my taking too many steps in one go and making too many assumptions...
The first image in post 33 (although the left/right labels on the posted image are the wrong way round) shows what happens if my 'TIF for UE2' last image in post 45 is plugged into UE2. This is a summary of the incorrect results:
n (front) - green (should be blue)
s (back) - magenta (should be yellow)
e (right) - yellow (should be cyan)
w (left) - blue (should be red)
u (top) - red (should be green)
d (bottom) - cyan (should be magenta)
I reasoned that if I went back to by original vertical cross image and swapped the colours of the spots accordingly, then went throgh the process of converting the resulting vertical cross to lat/long, flipping horizontally, convolving, saving as HDR, and running through omHDRconverter, then the resulting image, when applied to Light > Basic > Color of the UE2, should illuminate the cube and sphere correctly (i.e. blue light from the front, red from the left, green from the top, etc.
I think that step is logical, but I didn't actually try it out. Time to go and do that.
OK time to post something concrete
1°/ I started with generating a 1024x512 UV colored map from blender and tested it with blender cycle engine and it seems the IBL transformer is correcting everything OK
2°/ 1st render Mapped the map on default UE Sphere front view
3°/ 2nd render with IBL Transformer front view
4°/ 3rd render UE default envsphere Left view camera placed on ground
5°/ 4th render UE with IBL transformer Left view camera placed on ground
Seems there is not just azimut off for the envsphere. We can also see that the IBL transformer is applying a -1 scale to get the texture displayed correctly on the envsphere. These two errors make the default UE sphere completely useless. Yesterday I was scratching my head because I really can't understand how that is possible to provide a guide (the UE sphere) so much wrong
Here are four quick renders
First ist Blender Cycles with the UVgrid Color as environement
Second is UE with IBL transform
Third render is blender Cycle with another camera angle
Fourth render is DS with different camera angle
I get approximatively the same colors. I have yellow on the right side, pinkish on the left and cyan in front
Now for the Environnement Sphere Azimut off angle, I'm gonna setup a scene that will tell me what exactly the angle is
Back soon
I love this - three of us busy doing our own investigations along different paths... I do hope that we're all going to meet up at the same place eventually!
Horo:
Re post 37 ( http://www.daz3d.com/forums/discussion/28358/P30/#423353 )
I think the fact that IBL Transformer is necessary to get UE2 to work right (and there's at least three of us now who seem to believe that it does get it right) indicates that there is a solution and the solution is known. Though why IBL Transformer was released free but encrypted baffles me! A quick look through the script could've told us exactly what it does - but where's the fun in that? :cheese:
Is omnifreaker still around by any chance? Asking directly would be the most logical approach - but where's the fun in that, again? :cheese:
However, I'm now quietly confident that we're getting somewhere...
Takeo.Kensei:
Glad to hear that you've confirmed for youself that the IBL Transformer seems to get the UE2 working correctly. That's at least three of us (you, me and tofusan) in print on that matter, coming at it from different approaches. Unfortunately that's no use to Horo! But hopefully we'll end up with something that is.
Re the inverting of the image on the environment sphere. I'm fairly sure (from something I read on the UE2 wiki or one of omnifreaker's pages) that the UE2 environment sphere was actually intended to be a mirrored image - I didn't understand it then either! And I can't find the URL reference either. But like I said before, the environment sphere is a no-go area for me! Good luck with those investigations.
~~~~~~~~~~~
And now on with my own tests. The one step at a time approach works best for me - any time I try to take two steps in one go I fall flat on my face! So the next step is to see if can create a cubemap with my six coloured spots that, after my standard conversion process, will give me a TIF that gives the correct results in UE2 straight off.
First Image: My regenerated set of maps, with the coloured dots rearranged based on results of previous tests. The resulting TIF, plugged into UE2, should give me the correct illumination - i.e. each of the six lights appearing to come from the direction, and illuminating the face noted in post 45 http://www.daz3d.com/forums/discussion/28358/P30/#423995
Second image: Renders when applied to the UE2. Yaaay! The results are exactly as I'd expected.
Third image: Based on that, an original cubemap/environment arranged like this should work with UE2 (after flipping, convolving, saving as HDR, and omHDRconverterising of course).
Time to go away and test that now... and I think I've already spotted the mistake I made in the earlier post that I deleted!
I used an environment set originally created in Terragen Classic free. Everything's LDR (except that I export the convolved image from HDRShop as an HDR format so I can pump it through omHDRconverter. So the end result is an HDR TIF, but a pretty poor one.
Image 1: The original environment set. Sun is WSW, so it should illuminate the cube and sphere from back-left, slighty up.
Image 2: I went back to the original six cubemap renders and recreated everything based on the conclusion of my previous post.
Image 3: Plugged the resultant covolved TIF into UE2, upped the contrast to maximum (200%) to try and get a clue whether the sun illumination was coming from the right place, and rendered from both sides.
Sooooo... what to you folks think ???
(I'm off now - i need to soak my brain in a bowl of ice-cold water again! back tomorrow.)
Edit(9Sep): For my reference, just so I don't lose this! ImageMagick script to convert a standard (N at the centre of the cross, horizon horizontal) vertical cross cubemap (vc.jpg) to the rotated image 2 version (X.png is a 256x256 black square):
@3dcheapskate- Oh I do not doubt that IBL Transformer makes UE2 work correctly. Pity we do not know what exactly it does.
Two years ago when I reported the offset bug to DAZ 3D, I had asked for a contact to omnifreaker but I wasn't given it. I don't want to intrude on people but it would have probably helped to solve the issue(s) at the source.
@Takeo.Kensei - what! scale -1. I'm going to try this. I know that I can turn the inside out of a sphere and vice versa in Bryce but it didn't occur to me to try negative size in DS for the env-sphere.
Ah great. Size -100% flips the thing upside down but using either size X or Z at -100% does the trick - BUT only the environment sphere is flipped, the HDRI stays the same. Remember, the HDRI for the light and the env-sphere must be flipped.
When you rotate the UE2, both HDRI and env-map move in sync, as we would expect.
I know. There is an element I didn't post yet. The IBL Transformer really transforms the IBL
See by yourself
How can you make the HDRI visible in DS? Anyway, if the environment sphere AND the HDRI are flipped, everything is fine. This is what puzzles me. Only in UE2 can the env sphere be rotated or re-sized. It cannot be done separately. It makes sense that the two go together. But obviously not with size.
That would be another thing to verify. You know, in Bryce the HDRI is at infinite distance, like the sky, with which it can interact. Or it is a tiny sphere at world centre when used from inside (which is not natural, but comes in very handy for certain projects). The problem with a light sphere (the HDRI) at infinite distance is if you have an HDRI made indoors with lights quite near. Objects cast different shadows if illuminated from near or far. For outdoor HDRIs everything is fine. The sun, after all, is at infinite distance (well not really, but farther than the next light bulb). So if the diameter of the light sphere in UE2 can actually be set, that would be a really cool option.
No it's infinite from what I know
PS : I'm still scratching my head to find a scene which would give me the angle off of UE sphere but I guess I'll give up on that point. Not very easy and I don't think It brings any more information
Agreed, but I think you and Takeo are getting there - I'm not sure that I can really provide any further useful input, but I'll keep watching this thread with interest. (Personally speaking, simply knowing that setting UE2 X-Rotate and Y-Rotate to -90 makes the illumination look right is enough for my purposes. That lets me create TIFs for my environment sets that I can plug directly into UE2)
The attached image is simply for my reference - IBL Transformer loaded into DS4.5 with default settings (i.e. using its own test image), with the PNG version of the test image (there are also two TIFs - one visually identical, the other horizontally flipped) and the resulting render overlaid. There's definitely an explicit 18 degrees in there!
I know. There is an element I didn't post yet. The IBL Transformer really transforms the IBL
See by yourself
You've lost me there - I can't see anything that makes me think IBL Transformer transforms the IBL? But then again it's first thing in the morning here - this UE2 thing's been buzzing around in my head all night and I couldn't sleep properly.
Plus I haven't had my wake-up coffee yet, so my brain's not yet in gear...
You've lost me there - I can't see anything that makes me think IBL Transformer transforms the IBL? But then again it's first thing in the morning here - this UE2 thing's been buzzing around in my head all night and I couldn't sleep properly.
Plus I haven't had my wake-up coffee yet, so my brain's not yet in gear...
The IBL transformer does indeed transform the IBL.
If you picture the IBL as a UV mapped (light emitting) sphere it should roughly look similar to UE's environment sphere when it comes to how the image is mapped on the sphere. Of course this isn't the case, which is why we're here in the first place.
However the real problem with the IBL isn't that it's rotated in relation to the environment sphere but that the mapping is wrong. The "poles" of the IBL sphere should be on the top and bottom of the sphere were it mapped correctly to be used with equirectangular images but they are actually NOT. See the attached image for reference. The lighting (red from the top, turqoise from below) cast on the sphere primitive in the image should match how the UE environment sphere looks, but it clearly doesn't. The poles are wayy off from their proper locations.
The IBL transformer fixes this issue and you get a properly mapped IBL sphere which truly matches the environment sphere and how an equirectangular image actually should light the scene.
That does look interesting. Pity there's no way to make the HDRI directly visible. If the Transformer can correct all that, the flaws ought to be knowable. Then, HDRI and environment sphere could be made to compensate the errors.