White points in KE04
Hello,
What are these white points on the walls around the girl? they seem to come from reflexions.
I already see that in the past on very reflective surfaces (posted few months ago), but I didn't have the problem in the same image in KC23.
A bit after at 170 passes, more and more white points
Below is the rendering with KC23. No white points.
Views: 441
Replies to This Discussion
-
Permalink Reply by Render Plus on
-
Also, when I tested this same mofel the sun wasn't on. The sun is much brighter than the interior lights I ahd on in my tests.
If you still are getting white spots, save the model after the rendering starts and send it to me again, so that I am testing the same view and the same settings which you are having problems with.
-
Permalink Reply by Alain Bonnefoy on
-
I try to render an artificial lightning scene with KE26.
I got the setting from a saved setting file so, I suppose the setting is very close to the one I used with success.
Lights are the same, materials also excepting the floor.
Now, as you see, the result is a nightmare after close to 800 passes.
If Sun highlight is responsible to fuzzy sparkling, what about artificial lightning?
Thinking again about this subject, I have some difficulties to understand why we have to wait several hundreds of passes to remove points you intentionally put !?!
Regards,
-
-
I apologize for the trouble you are having. I know that artifacts like these dots can be frustrating.
It will help if you save a model which is failing and send it to me so I can see it here.
There can be problems with reflection of lights, but I need to have the model so I can see the settings.
Until I can test it myself, I can only guess what the problem may be.
Here is one possibility:
In order to process fuzzy (sharpness reflection not 1.0) reflection, we have to sample reflection from all directions. We only do a few of those per pass, so perhaps in the 123rd pass we will sample reflection from a direction which comes directly from the sun or a light on one pixel, but not for the next pixel, then the first pixel gets a white reflection and the next one does not. Then on later passes, this reflection is averaged into other directions.
This is just a guess, without being able to investigate the actual model. Almost all of our users don't have this problem. But the combination of lights, materials, reflection settings, and choice of rendering engine (Patch Tracer vs Packet Mode) may have created this problem for you.
-
-
Ok I prepare the model and setting file to send it to you.
I prefer to keep the model as is, even if it's large in case....
That night I reinstall KC23 to check rendering engine in same conditions. Below is the result:
So if few days ago I had a doubt about seeting used between KC23 and KE26, now it's almost sure there is a difference on the engine or the way the settings are used.
I send you an archive right now.
-
-
Thank you. We are looking at it now.
We will render one with the model you sent first to see if we get the same dots.
However, the back wall still has the fuzzy reflections.
I would:
A. turn off all reflection on the back wall - it isn't really necessary for this scene. That will probably solve the problem, but will not let us learn any more about it.
B. If you leave it on, run the sharpness up to 0.90 or even 1.00 to try to eliminate the dots. The low sharpmess setting it a prime culprit for these light source reflection artifacts.
C. Check the "No Light Source Reflections" check box. It was added to solve just the problem - it will have the back wall not try to directly reflect light sources. [Edit - This setting only works in Packet mode - it does not work with the Path Tracer]
I will try C first - sometimes, but not always it fixes this.
Then I will try B.
If that doesn't fix it, I will try A.
This model runs about 20 passes an hour on my machine, so it will take a while to get the results.
-
-
Also - our developer tells me that "Glossy" might be the setting we are looking for.
[Edit - This setting only works in Packet mode - it does not work with the Path Tracer]
Glossy Reflection does not reflect actual objects, but does have highlighting from lights.
Here is a teapot with Glossy Reflections (top) and normal (bottom).
-
-
After further reflection, (excuse the pun), we just need to render this more to see if it is a bug, or something else we can adjust for.
We shouldn't have to up the sharpness.
So give me 10 to 60 more hours to play with it, since the problem seems to show up only after a couple of hundred passes. I will see if I can duplicate it with with less geometry so I can test it more easily.
Al
-
-
I removed most of the geometry which was not visible, and rendered this with the Path Tracer for 100, 200, 400, 800, 1200, and 2000 passes. The results work mostly as expected. So this image doesn't tell us much - but it does show that the Path Tracer needs 1,000 to 2,000 passes to do a really good rendering.
This took about an hour. I am going to place the lady back in the model (Her clothes are also reflective) and see what effect that has. If that doesn't work, then I will render the original model. But I will need to let that run all night.
Also, after staring at this some I see that when I zoomed in - (so I could render a smaller area) I was not looking up at the ceiling as much. I will make that change as well.)
(These images are all blown up 2X to make any pixel artifacts more visible)
-
-
Adding the lady and lowering the camera didn't have any bad effects either.
(1600 passes with Path Tracer)
I will start a 1,000 pass render of the whole model before I leave for the weekend to see if I can duplicate your results.
I am going to send you this model, so you can try it there.
(I am rendering a zoomed in view at 400 pixel resolution and "Quick Draft" mode, so you should be able to get 1600 passes in about 2 hours.)
If it fails for you, save the .JPG (with comments) when it is done and send it to me. (We store all of our settings in the .JOG file, so I will be able to make sure that nothing changed)
If I still can't figure it out, I am going to need to get you to zip the entire nXt_Batch folder from your temporary directory and send it to me as well. (It contains everything we are doing in the rendering process. Some things - such as the exact view and size of the SketchUp window before rendering we cannot duplicate precisely when we exchange models)
Also, there is a new KF02 version on our downloads page. You should keep using these latest versions just in case something else we have fixed has fixed this problem as well.
-
-
Here is the full model rendering after 700 passes. We will see if it start to fail during the next 700.
(Quick Draft mode which doubles the size of each pixel)
-
-
Here is the result after 1200 passes. (I don't see any problems. ) This is a low res rendering exploded X2. I started one more higher res, zoomed in rendering to see if that effects the results.
If you get a similar problem again, with the latest version, zip up the entire folder %TEMP%\nxt_batch and send it so us.
If you send us the entire folder, we can try a rendering here with exactly the same view, same extract from SketchUp, etc. Hopefully we can duplicate the problem so we can try to fix it.
(Also, I hate to post these Quick Draft (2X) images - so I want to post a final image which shows the higher quality we get without Quick Draft.
-
-
Well ok Al
First, it seems you use path tracer engine. I used packet mode as it gives a less granular picture without to wait for a long time so, Can you check if the problem doesn't come from this mode?
Secondly, can you precise which version do you use?
I'm going to reinstall KE26 to render the model you sent to me.
Just a poiint, it seems we still have the problem about librairies not found on Windows 7 64 bits. Expecially when we try to choose an hdri sky.
Regards,
- ‹ Previous
- 1
- 2
- 3
- 4
- Next ›
About
Search
Enter a phrase here to search the entire Render Plus web site:
Loading
Translate
Latest Activity
© 2024 Created by Render Plus.
Powered by
