Ohhhh, thank you so much for this amazing comment. I really appreciate it! It's so nice to hear such kind words and that the work I do has such a good resonance!
Thank you very much. I doubt its the GOAT. I had to shave off so much stuff in order to squeeze it into the time frame that It's more like a light overview. But I am thankful for your great comment and kind words. Comments like yours make the whole endeavor worth it 🙌
Thank you very much for your nice and positive comment. It's true, there is a lot of info in there. But I could add at least info to 1h videos worth. So there is a lot that landed on the virtual cutting room floor. At least planning wise 😇 So great to hear your input and your positive reaction. Because I was thinking about doing smaller chunks, just e.g. talking a bit HDRIs for 15-20 minutes, but in more detail then covering a whole big topic in one video... We'll see how it goes. Would be interested to also hear your opinion about it if you would like to chime in.
@@SilverwingVFX Thank you. I would keep the maximum time not much further than 30 minutes per part, since it's a bit harder to watch an 1 hour video mentally than half an hour. So if you have nice segments of HDMI you could definitely divide those up into 20 minute chunks, easy to access and look up as well for when we need such info later. ( I always use a lot of your octane videos to return to when building stuff for reference) But in the end, we'll watch your content anyway since it's all very in depth and well presented, so as long as you yourself think you're doing a good job and it all fits nicely into your videos I think you'll be alright, you have a knack for finding the right content to show.
@@ThomvanVliet Thank you so much for taking the time to answer in such great detail and of course your input. Yeah I am absolutely with you on the "not longer then 30 Minutes" part. I think the best time is around 15 min. Not to long, not to short. I just almost never make it and those videos are ending up around 25 minutes mostly. But great to have a little bit more of an insight on your thoughts. Also great to hear that you think that I do a good job in providing interesting topics to explore! Cheers and thank you again for your insights 🙏🙌✨
Oh, super nice to hear! Though most things that I wrote down in my list for the 2nd part are rather self-explanatory... like in this vid. So I hope you can get something out of it 😇
I really enjoyed this video, it offers a clear explanation of how samples are calculated and why one scenario provides more noise than another. However, based on personal experience and conversations with fellow (non professional, or at least small time) Octane users in my circle, it seems that literal noise isn't the primary issue. Instead, fireflies appear to be rampant in Octane and what most users i have spoken with are actually meaning when they say noise, particularly in anything beyond the simplest scenes with minimal lights and hdri's. and they usually worsen with increased sample counts potentially completely engulfing the scene. This issue is particularly noticeable in architectural environments. To mitigate this, I often resort to employing numerous tiny emitters at the lowest possible values to ensure proper illumination of light fixtures. And then I utilize IES lights on invisible meshes or emission textures around fixtures to illuminate the room. Scenarios involving light bulbs within a specular medium around them, like those on ceiling fans, tend to exacerbate the occurrence of fireflies, often outnumbering the samples almost. Even when using exclusively IES or mesh lights (the ones that cause mush less fireflies than the worst case scenario), fireflies persist, especially when these lights are directed downward in close proximity to ceilings or walls. Placing lights in the center of the room results in zero fireflies, while positioning them less than a meter but pointing completely away from the ceiling results in fireflies. I've grappled with this issue throughout my year+ of learning Octane. Without discovering the Intel Open Image Denoiser and using it on octane image, fireflies would still pose a significant obstacle, rendering Octane results unusable as far as im concerned. I believe that a comprehensive video dedicated to eliminating fireflies would be a greatly beneficial video/ videos, and may actually be a larger part of the problem than actual noise. Blender octane user, greatly appreciative of your videos as they are my primary source for learning things octane over my journey. Thanks for the content.
Hey Fliguy1, thank you very much for your great and in detail comment. I appreciate you took the time to write it all down. I agree with you that Fireflies are part of the problem and due to their nature are much more prevelent then other noise. The really cool thing here is, that Fireflies are just a part of noise and therefore all the tricks I mention in this video at least reduce them. We will get into more goodies how to avoid them in the material and especially in the rendering part of the series. Fireflies are caused when a random indirect ray (unexpectedly) hits a very, very bright light source. They are either reflective or refractive caustics that are unresolved (meaning it would need more samples to even out the result) In the beginning of the video you can see what happens with the through the glass block caustics with a small light source when I make the light source bigger. You decrease the problem by 1: Making the light source more likely to find e.g bigger. 2 Through making the light source bigger, you also decrease its energy per area unit which also helps to avoid fireflies because the energy in the scene is more balanced (rays don't create as bright spots anymore because the lights in the scene are not as bright anymore, since the energy of the lights has been spread over a larger surface). Maybe I should make a special about fireflies and explain what they are directly. I decided to cut this video a bit short to fit it in to 20-30 minutes. So it's very distilled down to the essence and does not have a look at cases that could be made, as light portals ect. Spoiler for the rendering tutorial (which is rather well known anyway) Set the GI clamp down to 5-10. This lets indirect diffuse only carry up to a certain amount of brightness. So when the rays hit a really really bright light source, they only carry a fraction of it's intensity and therefore do not create a white spot (firefly) Obviously if you want to have nice caustics in your scene, then you would use a much higher GI clamp. Otherwise they will look dull. I thank you again for your comment and hope I could give you a little more insight into this. Cheers and a great mid of the week to you! Raphael
Amazing! Thank you. Really appreciate you take the time to explain with chart time. 😀 Your timing is great. Right in the middle of a project with lots of spot lights. 🔦
Hey Tristan, thank you very much for your great comment. Yaaaay ✨🔦✨ Nice to hear that I managed to produce something useful to you 🙌 I am just afraid that there is no good way to make spotlights sample better 🥲 But at least you have some of the other tips to go off of.
Hey hey and thank you very much. Yeah, It's pretty much universally applicable. Thanks for the Blender Octane Community for the input that made me do this video 🙏🙌✨
You are very welcome. I see you are working your way through my noise series and you are liking it. Thank you very much for all the nice comments you are leaving 🙌😊🙏
Thank you very much. The model for the intro was made in a CAD modeler. You might have heard of it, Plasticity. I wanted to try it since a lot of people were recommending it to me, and I have to say, it's really good indeed. SO this was a deviation from my usual Subdivision Surface modeling roots 🙌
You choose correctly the PT kernel/engine for highest realism level of lighting, if any glossy material with the roughness of less than 0.3~0.4 (eg: glossy ground) but the UNIQUE disadvantage of PT is the insane/super long number of samples per pixel (+/-10-million or longer at 1080p or +/-2- or 2.5-million or higher at 4320p) needed for good caustics at dot light and/or at our sunlight source(s) - if we use single GeForce RTX 4090, this may during +/-100 hours or longer time at full-image or at all regions picked. But fortunately, this kernel does not have any other nor second disadvantage. 🙂 At Photon Tracing kernel issue may not work render caustics for mix material but the Roeland (only?) developer can't fix this issue yet for few years. 😉
Always a bless! 🙌 I wish I had made it to FMX this year. I hope to see you there next year too. And I definitely need to grab one of these Otoy hoodies 🔥😁
Thank you very much ✨ I had the best FMX ever this year. But not because of the program or the Speakers. Because all the cool people I met there. If I have the resources, I will definitely will be there again next Year.
You seem to be not alone! Right now I have a couple of points but there are mainly the ones that are rather well known. Maybe I can talk to a couple of friends to get some more intel to take the next video from good to great 😇
I pretty much do the opposite of everything youre saying. Tiny Light Sources, and lots of them! Haha. Another tip Dino taught me was that you want to redue the amount of polygons in the mesh light, down to 1 triangle is the most optimized.
Ha ha ha. I think for your line of work, there is no other way then to do the opposite. And yes you are right about the mesh lights. If you can't avoid them all together, then its best to use as few polys as possible. I should have mentioned that while I was talking about not using mesh lights 😇
Thank you very much 🙏 The Denoiser will be part of the Render Settings Video. So yes. I think I will go over the settings. Though I think there is not that much to set there. Do you have any specific questions you want me to answer?
Thank you very much for your comment. Great to here you like it. I think you mean gobos? Basically it's rather easy. You use a texture in the distribution. Then you have to make the light rather small. The smaller the light the sharper the texture will appear. To control the texture a bit better in scale you can add a UV-Transform and use the scale there.
Ha ha ha. A white environment basically. It gives the renders their own unique style. Maybe a bit flat... but hey, it renders fast 😅😇 And thank you very much for the ✨🔦✨
It cannot be overstated the value of your contributions to this craft and community. Sincerest thanks for the efforts and knowledge you bring us.
Ohhhh, thank you so much for this amazing comment.
I really appreciate it! It's so nice to hear such kind words and that the work I do has such a good resonance!
ABSOLUTELY THE BEST DESCRIPTION EVER OF NOISE, SAMPLING AND PATHTRACING
Thank you sooo much.
Fantastic to hear that you liked the explanation 🙏😊✨
Finally! A very, very good (perhaps the GOAT) explanation of how the engines function. I will be so much better at optimizing scenes now. Thank you!
Thank you very much.
I doubt its the GOAT. I had to shave off so much stuff in order to squeeze it into the time frame that It's more like a light overview. But I am thankful for your great comment and kind words. Comments like yours make the whole endeavor worth it 🙌
Amazing lesson. Feels like a very in depth one but all within 30 minutes!
Thank you very much for your nice and positive comment.
It's true, there is a lot of info in there. But I could add at least info to 1h videos worth. So there is a lot that landed on the virtual cutting room floor. At least planning wise 😇
So great to hear your input and your positive reaction. Because I was thinking about doing smaller chunks, just e.g. talking a bit HDRIs for 15-20 minutes, but in more detail then covering a whole big topic in one video...
We'll see how it goes. Would be interested to also hear your opinion about it if you would like to chime in.
@@SilverwingVFX Thank you.
I would keep the maximum time not much further than 30 minutes per part, since it's a bit harder to watch an 1 hour video mentally than half an hour.
So if you have nice segments of HDMI you could definitely divide those up into 20 minute chunks, easy to access and look up as well for when we need such info later.
( I always use a lot of your octane videos to return to when building stuff for reference)
But in the end, we'll watch your content anyway since it's all very in depth and well presented, so as long as you yourself think you're doing a good job and it all fits nicely into your videos I think you'll be alright, you have a knack for finding the right content to show.
@@ThomvanVliet Thank you so much for taking the time to answer in such great detail and of course your input.
Yeah I am absolutely with you on the "not longer then 30 Minutes" part.
I think the best time is around 15 min. Not to long, not to short. I just almost never make it and those videos are ending up around 25 minutes mostly.
But great to have a little bit more of an insight on your thoughts.
Also great to hear that you think that I do a good job in providing interesting topics to explore!
Cheers and thank you again for your insights 🙏🙌✨
I can't wait for the second part thank you
Oh, super nice to hear!
Though most things that I wrote down in my list for the 2nd part are rather self-explanatory... like in this vid. So I hope you can get something out of it 😇
I really enjoyed this video, it offers a clear explanation of how samples are calculated and why one scenario provides more noise than another. However, based on personal experience and conversations with fellow (non professional, or at least small time) Octane users in my circle, it seems that literal noise isn't the primary issue. Instead, fireflies appear to be rampant in Octane and what most users i have spoken with are actually meaning when they say noise, particularly in anything beyond the simplest scenes with minimal lights and hdri's. and they usually worsen with increased sample counts potentially completely engulfing the scene. This issue is particularly noticeable in architectural environments.
To mitigate this, I often resort to employing numerous tiny emitters at the lowest possible values to ensure proper illumination of light fixtures. And then I utilize IES lights on invisible meshes or emission textures around fixtures to illuminate the room. Scenarios involving light bulbs within a specular medium around them, like those on ceiling fans, tend to exacerbate the occurrence of fireflies, often outnumbering the samples almost.
Even when using exclusively IES or mesh lights (the ones that cause mush less fireflies than the worst case scenario), fireflies persist, especially when these lights are directed downward in close proximity to ceilings or walls. Placing lights in the center of the room results in zero fireflies, while positioning them less than a meter but pointing completely away from the ceiling results in fireflies.
I've grappled with this issue throughout my year+ of learning Octane. Without discovering the Intel Open Image Denoiser and using it on octane image, fireflies would still pose a significant obstacle, rendering Octane results unusable as far as im concerned.
I believe that a comprehensive video dedicated to eliminating fireflies would be a greatly beneficial video/ videos, and may actually be a larger part of the problem than actual noise.
Blender octane user, greatly appreciative of your videos as they are my primary source for learning things octane over my journey. Thanks for the content.
Hey Fliguy1,
thank you very much for your great and in detail comment.
I appreciate you took the time to write it all down.
I agree with you that Fireflies are part of the problem and due to their nature are much more prevelent then other noise.
The really cool thing here is, that Fireflies are just a part of noise and therefore all the tricks I mention in this video at least reduce them.
We will get into more goodies how to avoid them in the material and especially in the rendering part of the series.
Fireflies are caused when a random indirect ray (unexpectedly) hits a very, very bright light source.
They are either reflective or refractive caustics that are unresolved (meaning it would need more samples to even out the result)
In the beginning of the video you can see what happens with the through the glass block caustics with a small light source when I make the light source bigger.
You decrease the problem by 1: Making the light source more likely to find e.g bigger. 2 Through making the light source bigger, you also decrease its energy per area unit which also helps to avoid fireflies because the energy in the scene is more balanced (rays don't create as bright spots anymore because the lights in the scene are not as bright anymore, since the energy of the lights has been spread over a larger surface).
Maybe I should make a special about fireflies and explain what they are directly.
I decided to cut this video a bit short to fit it in to 20-30 minutes. So it's very distilled down to the essence and does not have a look at cases that could be made, as light portals ect.
Spoiler for the rendering tutorial (which is rather well known anyway) Set the GI clamp down to 5-10. This lets indirect diffuse only carry up to a certain amount of brightness. So when the rays hit a really really bright light source, they only carry a fraction of it's intensity and therefore do not create a white spot (firefly) Obviously if you want to have nice caustics in your scene, then you would use a much higher GI clamp. Otherwise they will look dull.
I thank you again for your comment and hope I could give you a little more insight into this.
Cheers and a great mid of the week to you!
Raphael
Amazing! Thank you. Really appreciate you take the time to explain with chart time. 😀 Your timing is great. Right in the middle of a project with lots of spot lights. 🔦
Hey Tristan,
thank you very much for your great comment. Yaaaay ✨🔦✨
Nice to hear that I managed to produce something useful to you 🙌
I am just afraid that there is no good way to make spotlights sample better 🥲 But at least you have some of the other tips to go off of.
As always, very high quality useful informative content! Blender Octane users will be very grateful! :)
Hey hey and thank you very much.
Yeah, It's pretty much universally applicable. Thanks for the Blender Octane Community for the input that made me do this video 🙏🙌✨
thank you, it's so incredible😍
You are very welcome. I see you are working your way through my noise series and you are liking it. Thank you very much for all the nice comments you are leaving 🙌😊🙏
@@SilverwingVFX I watched it and it was great, thanks again
🔦 It is so helpful. Thank you!
Thank you very much. Glad to hear that 🙌
Awesome tut! Btw, your title animation is so beautiful! Curious about how you make it especially the inside structure of the plastic!
Thank you very much. The model for the intro was made in a CAD modeler. You might have heard of it, Plasticity.
I wanted to try it since a lot of people were recommending it to me, and I have to say, it's really good indeed.
SO this was a deviation from my usual Subdivision Surface modeling roots 🙌
You choose correctly the PT kernel/engine for highest realism level of lighting, if any glossy material with the roughness of less than 0.3~0.4 (eg: glossy ground) but the UNIQUE disadvantage of PT is the insane/super long number of samples per pixel (+/-10-million or longer at 1080p or +/-2- or 2.5-million or higher at 4320p) needed for good caustics at dot light and/or at our sunlight source(s) - if we use single GeForce RTX 4090, this may during +/-100 hours or longer time at full-image or at all regions picked. But fortunately, this kernel does not have any other nor second disadvantage. 🙂 At Photon Tracing kernel issue may not work render caustics for mix material but the Roeland (only?) developer can't fix this issue yet for few years. 😉
Great video!
Thank you very much 🙌🙌
🔦 Thank you for the other great video. I look forward to the next one!
Also thank you very much for your nice comment and of course for watching. Really appreciate it ✨🔦✨
Always a bless! 🙌
I wish I had made it to FMX this year. I hope to see you there next year too. And I definitely need to grab one of these Otoy hoodies 🔥😁
Thank you very much ✨
I had the best FMX ever this year. But not because of the program or the Speakers. Because all the cool people I met there. If I have the resources, I will definitely will be there again next Year.
thanks you for the tips, and the good vibes !
Thank you also for giving me good vibes through your comment 🙌
Waiting for the next part, Happy Noise Scrubbing!
You seem to be not alone!
Right now I have a couple of points but there are mainly the ones that are rather well known.
Maybe I can talk to a couple of friends to get some more intel to take the next video from good to great 😇
Ohhhhhhhh, that mind-blowing ❤ thank you soooooo much
Glad you like it that much. Much appreciated 🙏 🙌 😊
I pretty much do the opposite of everything youre saying. Tiny Light Sources, and lots of them! Haha. Another tip Dino taught me was that you want to redue the amount of polygons in the mesh light, down to 1 triangle is the most optimized.
Ha ha ha. I think for your line of work, there is no other way then to do the opposite.
And yes you are right about the mesh lights.
If you can't avoid them all together, then its best to use as few polys as possible. I should have mentioned that while I was talking about not using mesh lights 😇
Great, as always. Any chance you do a tutorial on denoiser settings? Thanks!
Thank you very much 🙏
The Denoiser will be part of the Render Settings Video. So yes. I think I will go over the settings. Though I think there is not that much to set there. Do you have any specific questions you want me to answer?
@@SilverwingVFX Well, is there a way to inlfuence the amount of denoiser or is it ON or OFF?
Always quality !
Thank you. I'm trying 🙏🙌
Thank you for sharing.
You are very welcome! Thank you for watching ✨🔦✨
Thank you for this awesome depth tutorial!! 🔥 Is it possible to request a tutorial about hobos with path tracing? Thank you so much
Thank you very much for your comment. Great to here you like it.
I think you mean gobos?
Basically it's rather easy. You use a texture in the distribution. Then you have to make the light rather small. The smaller the light the sharper the texture will appear.
To control the texture a bit better in scale you can add a UV-Transform and use the scale there.
@@SilverwingVFX Thanks for the reply!! i'lll gor try it out! thanks Silver! u the boss
Thank you!
And also thank you ✨🙌✨
Спасбо!❤
Ласкаво просимо 🙌
Muy bien gracias
De nada
¡Gracias por tu comentario!
🔦 For now on, I will fight against noise with bigger lights. I will create the biggest lights ever.
Ha ha ha. A white environment basically. It gives the renders their own unique style. Maybe a bit flat... but hey, it renders fast 😅😇 And thank you very much for the ✨🔦✨
I use vray but i still enjoy the videos 🔦
Man, that's a super huge compliment. I really appreciate that you like the videos and watch them even to the end ✨🔦✨ 🥰
🔦
✨🔦✨
🔦 this flash light has important sampling turned OFF . ❤
😁😆 Ha ha ha. It will be a very noisy very noisy flashlight then. ✨🔦✨
🔦
Ha ha nice. The daily token of appreciation from you. Muchos gracias 🙌🔦🙌