Loading...
> >

September 2024 - Daz 3D New User Challenge - Depth of Field and Canvases

2»

Comments

  • iaunpomiaunpom Posts: 34

    This is another topic, but after looking at, for example, the emissive surfaces in this image, the candle and the flame, which is an emissive surface that is a little bit in the background of the image with a little bit of DOF. Iray puts some pixels around the emissive surface that I don't understand. Maybe it's that the emissive maps of the flame are very few pixels and I would have to expand the UV mapping, but these pixels remind me of what happens when you make a depth map with canvases, without having the entire camera frame full of mesh. If in the background frame you see an HDRI of the global illumination and you don't put an object that covers the entire background frame as a plane, when making the canvas depth map, the mesh objects have some pixels on the edges, which Iray can't place due to lack of reference and can't do the anti-aliasing. That has made me think that there is a similar situation with the emissive surfaces, I don't know if it's something related to the convergence of the rendering, I haven't tried the degrain noise in these cases or the post denoiser, correction, I just looked at some render I made of the scene with denoiser and logically those pixels are almost not seen.

    Would I have to render on canvases with the post denoiser these types of surfaces separately ? If you rendered it separately, what happens with those few pixels, in the sample image, would I say that those pixels are also seen in the reflection of the glass and not beyond the candle recipe maker throughout the scene?

    Best regards

    Ejemplo.jpg
    1416 x 969 - 344K
  • LinwellyLinwelly Posts: 5,914

    iaunpom said:

    This is another topic, but after looking at, for example, the emissive surfaces in this image, the candle and the flame, which is an emissive surface that is a little bit in the background of the image with a little bit of DOF. Iray puts some pixels around the emissive surface that I don't understand. Maybe it's that the emissive maps of the flame are very few pixels and I would have to expand the UV mapping, but these pixels remind me of what happens when you make a depth map with canvases, without having the entire camera frame full of mesh. If in the background frame you see an HDRI of the global illumination and you don't put an object that covers the entire background frame as a plane, when making the canvas depth map, the mesh objects have some pixels on the edges, which Iray can't place due to lack of reference and can't do the anti-aliasing. That has made me think that there is a similar situation with the emissive surfaces, I don't know if it's something related to the convergence of the rendering, I haven't tried the degrain noise in these cases or the post denoiser, correction, I just looked at some render I made of the scene with denoiser and logically those pixels are almost not seen.

    Would I have to render on canvases with the post denoiser these types of surfaces separately ? If you rendered it separately, what happens with those few pixels, in the sample image, would I say that those pixels are also seen in the reflection of the glass and not beyond the candle recipe maker throughout the scene?

    Best regards

    this looks very much like it wasn't rendered for long enough, which can happen in rather dark images. there are several things you can do in this case:

    in the Progressive Rendering settings you can dial up the Max Time and the Max Samples for the image to render longer and in consequence have a better quality to work with whatever you plan after that (using canveses for example)

    in the filter settings I always advice to set the Pixel filter radius to 1 or lower (the standard is 1.5, which results in faster but slightly blurry renders)

    Also set the firefly filter enable to on

    Personally I'm not a fan of the Noise Degrain Filter and have that set to 0. I rather experiment with the settings until I have a clear render and in the rare case it doesn't work out I prefer to take my images into a postwork image editing program. But this is my personal experience and preference, and you need to find there what works best for you.

    In most cases it also helps to make the whole scene brighter and darken it in postwork in order to get rid of stray pixels

  • iaunpomiaunpom Posts: 34

    Linwelly said:

    iaunpom said:

    iaunpom said:

    I thought that maybe there was a relationship when using the DOF, after reading over some of the references, for example when putting an aperture in tone mapping with more light, then in the DOF for example, the areas with light are sharper or the focus areas of the DOF have a higher exposure.

    This comment after a few days seems a bit confusing to me. I had a theory that if you use the DOF, the most out-of-focus areas would have to be less bright than the sharper ones. For example, if you have a light source that is not directly seen on the screen but is seen on a reflective surface such as a mirror or other highly reflective surface and these surfaces are in the out-of-focus area, my question was or is it what happens to the luminosity of the light source, it adds to that of the surface where it is reflected and the nearby surfaces in blur,  or the surrounding blurred surfaces detract from the blur of the light source.

     

     I'm still unsure what the question is here, so let us try to disect it a bit

    1. the standart f/stop in the tone map settings is 8. if you set it to 16 the amount of light that reaches the camera is reduced, if you set it to 4, more light reaches the camera (directly connected with the shutter speed if you rule out the exposure value)

    2. if you want an image with sharp contrasts, you want to set your tone mapping f/stop to a hight value (eg 16) but you need more light in your scene for that to work (in the case of using the sun sky rest the SS multiplier from 0.1 to 0.5 or somethign similar) otherwise your scene becomes a muddy grey

    3. none of that has a direct connection to the camera f/stop in DAZ (this is different with a real life camera) a low value in f/stop with give you a very small area in focus in your scene, a high value will make the focus area larger. This range changes as well with the distance of the camera to the focus point the farther away the camera, the larger the area. As well the focal length will affect the area of focus in your image. To make this more visible, try to look at your camera in the scene through the perspective view and see how the focus lines move when you change the settings.

    I hope this answers your question? Again the best way to understand is to try several settings in a simple scene that doesn't take too long to render.

     

     I try to have a broader understanding of the rendering mechanisms and how to render each element. Whether to do it by separate canvases or parts or a rendering of the entire scene. As they say, each situation, depending on the elements, gives some results or others.
    Simplifying, I try to understand how Iray interprets the situation of: you have a scene in which you see the sun, for example, a person if he looks directly at the sun is dazzled and if he does not look directly at the sun he sees the entire field of vision, including the sun. If you render for example a similar scene, how should it look?, if you add the bloom filter even if you are not focusing directly on the sun you can create the glare effect. And in outdoor scenes where the sun is not directly on the scene but falls directly on some parts or with emissive surfaces, the configuration of the Bloom filter seems complicated to me at the moment.  

  • iaunpomiaunpom Posts: 34

    Linwelly said:

    iaunpom said:

    This is another topic, but after looking at, for example, the emissive surfaces in this image, the candle and the flame, which is an emissive surface that is a little bit in the background of the image with a little bit of DOF. Iray puts some pixels around the emissive surface that I don't understand. Maybe it's that the emissive maps of the flame are very few pixels and I would have to expand the UV mapping, but these pixels remind me of what happens when you make a depth map with canvases, without having the entire camera frame full of mesh. If in the background frame you see an HDRI of the global illumination and you don't put an object that covers the entire background frame as a plane, when making the canvas depth map, the mesh objects have some pixels on the edges, which Iray can't place due to lack of reference and can't do the anti-aliasing. That has made me think that there is a similar situation with the emissive surfaces, I don't know if it's something related to the convergence of the rendering, I haven't tried the degrain noise in these cases or the post denoiser, correction, I just looked at some render I made of the scene with denoiser and logically those pixels are almost not seen.

    Would I have to render on canvases with the post denoiser these types of surfaces separately ? If you rendered it separately, what happens with those few pixels, in the sample image, would I say that those pixels are also seen in the reflection of the glass and not beyond the candle recipe maker throughout the scene?

    Best regards

    this looks very much like it wasn't rendered for long enough, which can happen in rather dark images. there are several things you can do in this case:

    in the Progressive Rendering settings you can dial up the Max Time and the Max Samples for the image to render longer and in consequence have a better quality to work with whatever you plan after that (using canveses for example)

    in the filter settings I always advice to set the Pixel filter radius to 1 or lower (the standard is 1.5, which results in faster but slightly blurry renders)

    Also set the firefly filter enable to on

    Personally I'm not a fan of the Noise Degrain Filter and have that set to 0. I rather experiment with the settings until I have a clear render and in the rare case it doesn't work out I prefer to take my images into a postwork image editing program. But this is my personal experience and preference, and you need to find there what works best for you.

    In most cases it also helps to make the whole scene brighter and darken it in postwork in order to get rid of stray pixels

     

    Yes, the rendering is short, but the three four pixels are still in renders of more pixels, reflecting and expanding caustically in the room, I would say. And the scene has lighting difficulties, looking for realistic night lighting and for the moon to illuminate part of the scene with the Sun-Sky, using the sun as the moon ... I'm still thinking about how to solve it... Where does the sun go using Daz's Sun-Sky when it crosses the horizon line ? Just kidding, it's not a question, just a thought I have, but... 

  • FishtalesFishtales Posts: 6,102

    For Moonlight in a night scene outside I use a distant light set at 4500 temperature, 0.5 Lumen then set up the Tone Mapping to get the brightness. I use camera settings which can work out at Shutter Speed down to 30, and ISO up to 800.

    Playing about with the Australia Ecology, DOF and lighting (just Sun and Sky). You can se the lights on the Tank are just round circles but before that they were larger and full of dots which gradually disappeared as the render finished. A Denoiser wasn't used. The sharp area is across the front of the rocket in the launcher.

    No DOF in this one.

    2024-09-13 15:27:44.173 [INFO] :: Total Rendering Time: 47 minutes 28.97 seconds

    2024-09-14 12:47:54.811 [INFO] :: Total Rendering Time: 59 minutes 40.9 seconds

    2024-09-15 11:23:46.630 [INFO] :: Total Rendering Time: 40 minutes 10.75 seconds

    2024-09-15 11:54:20.872 [INFO] :: Total Rendering Time: 18 minutes 40.78 seconds

    2024-09-15 12:23:02.097 [INFO] :: Total Rendering Time: 10 minutes 48.50 seconds

    Click on images for full size.

  • LinwellyLinwelly Posts: 5,914

    iaunpom said:

    Linwelly said:

    iaunpom said:

    This is another topic, but after looking at, for example, the emissive surfaces in this image, the candle and the flame, which is an emissive surface that is a little bit in the background of the image with a little bit of DOF. Iray puts some pixels around the emissive surface that I don't understand. Maybe it's that the emissive maps of the flame are very few pixels and I would have to expand the UV mapping, but these pixels remind me of what happens when you make a depth map with canvases, without having the entire camera frame full of mesh. If in the background frame you see an HDRI of the global illumination and you don't put an object that covers the entire background frame as a plane, when making the canvas depth map, the mesh objects have some pixels on the edges, which Iray can't place due to lack of reference and can't do the anti-aliasing. That has made me think that there is a similar situation with the emissive surfaces, I don't know if it's something related to the convergence of the rendering, I haven't tried the degrain noise in these cases or the post denoiser, correction, I just looked at some render I made of the scene with denoiser and logically those pixels are almost not seen.

    Would I have to render on canvases with the post denoiser these types of surfaces separately ? If you rendered it separately, what happens with those few pixels, in the sample image, would I say that those pixels are also seen in the reflection of the glass and not beyond the candle recipe maker throughout the scene?

    Best regards

    this looks very much like it wasn't rendered for long enough, which can happen in rather dark images. there are several things you can do in this case:

    in the Progressive Rendering settings you can dial up the Max Time and the Max Samples for the image to render longer and in consequence have a better quality to work with whatever you plan after that (using canveses for example)

    in the filter settings I always advice to set the Pixel filter radius to 1 or lower (the standard is 1.5, which results in faster but slightly blurry renders)

    Also set the firefly filter enable to on

    Personally I'm not a fan of the Noise Degrain Filter and have that set to 0. I rather experiment with the settings until I have a clear render and in the rare case it doesn't work out I prefer to take my images into a postwork image editing program. But this is my personal experience and preference, and you need to find there what works best for you.

    In most cases it also helps to make the whole scene brighter and darken it in postwork in order to get rid of stray pixels

     

    Yes, the rendering is short, but the three four pixels are still in renders of more pixels, reflecting and expanding caustically in the room, I would say. And the scene has lighting difficulties, looking for realistic night lighting and for the moon to illuminate part of the scene with the Sun-Sky, using the sun as the moon ... I'm still thinking about how to solve it... Where does the sun go using Daz's Sun-Sky when it crosses the horizon line ? Just kidding, it's not a question, just a thought I have, but... 

    it might be easyest if you make a screenshot of your settings together with the full scene rendered, so I can take a look what is really there and what might have caused it

  • iaunpomiaunpom Posts: 34

    This is an image with 5000 samples, the first one I have sent I could not say, but it would be about 1000 samples. In the 5000 it seems that those pixels are starting to be integrated more into the image, I don't know how many samples I would need to finish those pixels fully integrated. I've been varying the rendering settings, what I think affects the pixels the most maybe is the spectral rendering option that I don't understand very well, but it does like a background of colored pixels that are integrated and mixed to make something similar to a scattering effect of the colors of the surfaces. When you look at those more blended pixels of the flame it looks like they're just blending with the RGB primary colors.

    I also have to look at the resolution of the UV mapping of the flame and zoom in to see if there is any variation, I don't know what kind of effect it has on those pixels but they seem like few pixels to me. It has a 2048x2048 texture map and this is the mapping.

    The original set I used doesn't have any mapping on the emissive surface of the flame and I've made one to try to do some details, that may also cause those pixels and it's not originally intended to be mapped.

    I'll try on a simple set I have from the scene, the candle from the original set and the one I've modified together, to see what the result is

    5000_Samples.jpg
    671 x 794 - 197K
    Mapping UV.jpg
    789 x 792 - 249K
    Texture.jpg
    672 x 551 - 39K
    Render_1000_Samples.jpg
    1890 x 1080 - 2M
  • FishtalesFishtales Posts: 6,102

    The coloured pixels around the flame are from the refraction through the glass cover.

  • iaunpomiaunpom Posts: 34

    Fishtales said:

    The coloured pixels around the flame are from the refraction through the glass cover.

    It's a posibility

     

  • iaunpomiaunpom Posts: 34
    edited September 17

    But no, it seems to be something more related to the difference in white or luminosity values of the pixels, because the scene has dim light and the luminosity value of the pixels around the candle flame is very distant and generates convergence difficulties for Iray, I will ask in the nuts and bolts forum to see if there is any solution to solve this type of rendering situations.

    In the test images with the original object it seems to make fewer pixels but also the flame is smaller, I will try to make the UV map of the flame bigger.

    Whitout_Glass_cover.jpg
    1620 x 1620 - 1008K
    1000_Samples copia.jpg
    1620 x 1620 - 696K
    1000_Samples_Spectral rendering ON copia.jpg
    1620 x 1620 - 1M
    Post edited by iaunpom on
  • LinwellyLinwelly Posts: 5,914

    iaunpom said:

    But no, it seems to be something more related to the difference in white or luminosity values of the pixels, because the scene has dim light and the luminosity value of the pixels around the candle flame is very distant and generates convergence difficulties for Iray, I will ask in the nuts and bolts forum to see if there is any solution to solve this type of rendering situations.

    In the test images with the original object it seems to make fewer pixels but also the flame is smaller, I will try to make the UV map of the flame bigger.

    can you tell me which product you used for the candle and flame, so I can try and test it, and see what's going on?

  • iaunpomiaunpom Posts: 34

    Yes, Rustic Living Room Product ID: 24082. And it's not just the candle, the emissive surface of the fireplace does too, I had a canvas rendering and I've noticed. I'm going to do some renderings to try to explain what it said.

    Emissive canvase.jpg
    972 x 788 - 229K
  • LinwellyLinwelly Posts: 5,914

    iaunpom said:

    Yes, Rustic Living Room Product ID: 24082. And it's not just the candle, the emissive surface of the fireplace does too, I had a canvas rendering and I've noticed. I'm going to do some renderings to try to explain what it said.

    I own that product, will look into it

  • LinwellyLinwelly Posts: 5,914
    edited September 18

    so @iaunpom can you do me a favour and test render that scene with the firefly filter off? (default is on so I assume you had it probably on all the time)

    here are some experiments I did

    this one is the room area of the fire and candles rendered without any changes (dome scene standart)

    and this is a closeup on the candle and yes I get those stray pixeld as well (not fully convergenced but that would have taken too long seeing that it's only at 8% at that point)

    here is the mesh of the candleflame and since this is a funny egg with all the surface being emissive there shouldn't be any transitional pixels which is different for the fireplace that has a transmap that defines the amount of luminance besides the emissive map

    for a test I turned off the firefly filter and was very surprised that this results in a much brighter overall image

    so I tried with the camera back to the complete fireplace with candles and those were much brighter there as well.

    of course now we have stray light pixels all over the place now but the amount by which the overall light setting gets reduced is really baflling me

    both the first and the last were rendered to the standard 5000 Samples. More smaples would have gotten rid a a substancial amount of stray pixels

     

    So I would like to know if you get similar results for this situation, as well I will ask back about that difference the firefly filter makes

     

    rusticliving_nochanges.png
    1362 x 912 - 1M
    rusticliving_candleclose.png
    1276 x 828 - 804K
    rusticliving_candlemesh.png
    1438 x 870 - 252K
    rusticliving_candlefrireflyoff.png
    920 x 584 - 221K
    rusticliving_fireflyoff.png
    1366 x 904 - 1M
    Post edited by Linwelly on
  • iaunpomiaunpom Posts: 34

    About the firefly filter the difference in ON and OFF in the preview at 200 samples I get more white pixels if I look at the candle.  Img_1 ON Img_1 OFF

    With the Post Denoiser ON filter, the results that appear in the preview are similar to those you have shown me with the firefly filter in ON and OFF. Img_2 ON Img_2 OFF

    With the Caustic Sampler ON in the preview to 200 samples and without firefly filter or post Denoiser the result that appears to me. Img_3 ON

     

    And if I leave aside how tedious it is for me to find the contrasted technical information of the result of those pixels in the rendering, now I know that candles and those pixels do not generate reflection or caustic refraction in the room. And the set-up generates a beautiful, perhaps casual, poetics. When activating the Bloom filter on the last setting Img_3 ON, it seems that perhaps the candles are containers that were full of fireflies, or it is what a room in the field has and they have entered through the window. 
    Img_4 

    By activating the firefly filter in the Img_4 setting the fireflies leave. Img_4 ON

    It is an illusion, the fireflies in the room are generated by the lamp that is out of frame, when hiding or extinguishing it the fireflies leave and are not generated by the candles or the glass container with the set configuration. Img_5


    Images Img_4, Img_4 ON, Img_5. They have another lighting setup I've closed the scene and I couldn't say for sure, but I think it's this one. Img_6.

     

    Img_1_ON.jpg
    1494 x 961 - 859K
    Img_1_OFF.jpg
    1503 x 966 - 933K
    Img2_OFF.jpg
    1498 x 951 - 399K
    Img2_OFF.jpg
    1498 x 951 - 399K
    Img_3_ON.jpg
    1496 x 962 - 962K
    Img_4.jpg
    1495 x 963 - 1M
    Img_4_ON.jpg
    1495 x 950 - 1M
    Img_5.jpg
    1492 x 958 - 763K
    Img_6.jpg
    455 x 482 - 136K
  • LinwellyLinwelly Posts: 5,914

    @iaunpom thanks a lot for testing all that, at this point the stray pixels you find around the emissive light sources seem to be a technical problem, that will not be solved that easily, so the question is if I can help you find a way to work around it.  What is the scene you are planning on with that set? For example if you wanted to place a character or something else for a story in front of the fireplace you could use DOF to focus on the character and with that get rid of stray pixels in a more general way,

    or you could just go and replace the candles with some other candles which are transmapped

    rustic-character.png
    800 x 600 - 832K
Sign In or Register to comment.