Dude this is literally genius you're amazing! I always had hacky ways of dealing with double shadows and have never seen such a good method for dealing with this. Thank you so much
Thank you so much. Your tutorials are very good. Please could you make a tutorial on the noise node and it's goodness, cause I'm trying to understand that node very well. I know it's very powerful
for crossing shadows I use a constant/voronoi gradient to fake the real one and min this with the cg mask as mask input. The rest of implementation is neat though
The way how you create your shadow in this complex case is really great; "min"-ing a shadow can help in a lot of simpler cases to add the correct shadow in a quick and efficient way, though and definitely without bothering "stencil"-ing the shadows or paint them away
Sorry if i'm asking something after too much time. I'm working on a shot, and I'm trying to create a "shadow cleanplate" similare to yours, but when I apply the keyer effect directly to the colorspace node it doesn't work as expected, it just make darker the dark areas. I get something similar deactivating the red and green colours on the node
Nice tutorial, I am kind of confused about the part where you set the colorspace to HSV but then used a regular luminance keyer to key the highlights in HSV. Should not the keyer be set to blue keyer, or am I missing something?
You can set it to the blue keyer, that would be correct. In my script here, I copied back in the Red and Green from the original, which is doing the same thing but in a different way.
Hey Alexander, Thanks for you tutorial. Very interesting way of adding shadows. This reminded me of a tutorial from FXPHD Tips and Tricks Volume 3, class 05. This is also about adding shadows and this technique involves the exposure node. I would love to know your take on that. Probably your way matches the attenuation of the shadows more.
Thanks! I haven't seen that particular tutorial so I can't really compare. But it sounds like they're just using an exposure node instead of multiply which basically do the same thing, except that the exposure node is measured in camera values ("stops" of light), vs just 0-1 like a normal grade. Visually though there's no difference.
Thank you very much for posting this- great information here! I was curious on another topic. If you're using a Nuke spotlight in a 3D scene, how do you get the spotlight to create soft cast shadows like you'd see in nature? So far all I've been able to get are hard shadows with the spot or directional light. Thanks in advance!
Unfortunately there's no easy way to do a soft shadow in Nuke as far as I'm aware. There's iBlur which can simulate a soft shadow easily, but if you wanted to just cast soft shadows you would probably need to build a gizmo that does this. I have a few ideas on how you could probably build that tool, but it would be hard to explain on a text. Hopefully though they just add some more features to Rayrender so we can get soft shadows in a less confusing way.
This is a neat trick. I am staring at that HSV image though and I'm noticing the shadow is purple but the floor is blue... Isn't there a From merge node trick in there somewhere that could be used with that colourspace? i.e. painting the tree shadow to blue first and using the difference info somehow?
You definitely could. What you're seeing primarily is difference in hue which is why it's appearing more red. Basically the warmer areas / areas in the light are warmer in hue, so you could do similar to what you said or just do some color grades which might be faster.
You probably could do a lower number and do more “steps”, personally I haven’t tested it. It would mean a lot of nodes, but generally 0.5 and just copying hasn’t produced any obvious color banding for me in most cases. Worth testing out though, might give a good result!
I’ve been doing vfx since I was young, but also had few great instructors in college (full sail). The primary instructor I had no longer works there. However a lot of the advanced stuff I learned mostly from working with other people in vfx studios, as well as with VFX supervisors at places like ILM or Weta where they have 20+ years of experience. Some of those guys have incredible artistic eyes and technical understanding, so when you spend a lot of time with them you learn exponentially
@@CompositingAcademy Holy cow you worked at ILM and Weta. That is a very proud moment Sir. As a person living in third world country, I can't study VFX because of no VFX college. That's why UA-cam and your courses helped me out. But I don't know how would I learn these things from scratch or land a job. Thanks for replying me.
Awesome I’m glad they have been useful! My goal is to make a competitor to vfx colleges because they’re vastly overpriced and many students leave with an insane amount of debt. I believe I can put something together that will be better than those programs at a fraction of the cost, so I’m working hard on that this year! The courses were structured in order for that reason, but I will be adding some additional elements to finalize the whole package as well as having live teaching included for those who want it.
@@CompositingAcademy The debt part is pretty scary although I agreed that learning VFX is pretty expensive. You are doing pretty good with the quality of NUKE series specially NUKE 707 has Transformers like CGI quality. I think you are not just a compositer but also a great CG Generalist because I remember you modeled porsche in NUKE 303 which was fantastic. The effort you are putting in NUKE series are astonishing. I have a small question and an obstacle which is bugging me alot. Like compositer role is to get all final results from other artists in order to complete his work but how would I make a project or showreel because I can't add the assignments I did from courses in my reel. I hope you are getting my point because how would I tackle other roles in it or should I collab with other students from different field for making a reel. Thanks in advance.
Hey Edmund, The Udemy beginner courses have english subtitles, but not multi-language. However, if you google for Udemy subtitle translate, there are ways to do it. Cheers!
The main way would be figuring out how to get a good shadow cleanplate, and then just blending that into the shadow alpha. You can roll off the highlights in a similar way probably by using the Curves tool in After Effects, and just repeating that effect until you have a similar 'flat' image. Would have to play around with it as I haven't used after effects in a few years, but the principle is pretty much exactly the same: target the highlights, bring them down a bit, and keep doing that over and over to get a shadow plate.
Thanks! I think HSV and HSL are pretty much the same, the math is slightly different. Ben Mcewan wrote on his blog that explains the difference pretty well: "HSV is the same as HSL, except the V stands for “Value”. There is a difference between luminance and value based off the math used, but it’s a similar concept. If you multiply the blue channel in HSL vs HSV colourspace (blue channel, because Blue has been converted to Value), you’ll notice that HSV produces a similar result to just multiplying your image the regular way, whereas HSL appears to desaturate the highlights as they get brighter. HSV also seems to give a flatter, more diffuse look to the saturation values, so you’d probably want to use HSV over HSL in most cases."
@@CompositingAcademy thanks for the reply. I often use HSL. Going start to use HSV more times now. Besides that, one thing came to my mind: this "isolation process" (let's call this way) by changing the colour space then changing back to the original, shouldn't isolate hue saturation and Luma (or value) information in each isolate channel R,G and B? so why this difference between HSV and HSL affects the saturation since the channel for that is the green one? does it make sense?
@@elironerosa3609 Hey! Sorry didn't see this comment before. I'm not entirely sure to be honest, I think the way the "Luminance" channel is created must be slightly different, which is why it is affecting the saturation slightly when using that alternative option. My guess is that it's doing some math behind the scenes on the R G B channels, but these two colorspaces do some slightly different 'difference' operations to create that brightness channel. In other words, it might be "weighting" one channel more than the other when it goes into the highlights, when calculating what is the luminance vs color. Probably someone who knows further in color science or the guys who work at foundry could answer better than that, but that's about what I know!
Has anyone been able to get this tactic to work? On my end it doesn't really do much to highlights, and just makes the image WAY darker without revealing/leaving the details behind. Exact same node setup. When googling, looks like another person had same issue.... working in linear etc.... !
hmm usually it works, if your highlights are completely clamped and there's no more detail in them - then there's no way to bring in more detail without painting the detail in. If your footage has some dynamic range it should work. If there's detail in the brighter areas like the example image ground area, it should also work.
that was rendered in lighting. I had a directional light for the sun, then I just used a geometry to block the sunlight where the tree is placed, to cast a shadow. Tricky to get the angle lined up with the real shadows but it works
@@belovedangels6457 I believe I used light linking / shadow linking. So that the tree shadow would only cast onto the cube but not the ground as to not create a duplicate shadow. Then the cube would still cast a shadow. Been a while since I did this tutorial, but if it's not that then the other way would be to just render the shadow pass separate (disable the trees and only render the box shadow).
This method works on video as well. You would just framehold the video, create the shadow cleanplate from the still image (the same as explained here), and project that out onto 3d ground geometry.
Hey! Most definitely you could, in After Effects or Fusion probably would be the easiest. I'm not too familiar with Blender's compositor, but it probably could do this. Honestly though it's best to learn either Nuke or Fusion for more advanced compositing. Nuke has a non commercial version that is free to use for personal projects!
![Re-lighting Real Footage | Nuke Compositing [Advanced]](http://i.ytimg.com/vi/VYjmvB6d9NA/mqdefault.jpg)
![Re-lighting Real Footage | Nuke Compositing [Advanced]](/img/tr.png)
![Compositing in UV space with Projections | Nuke [Advanced]](http://i.ytimg.com/vi/F-q8tgk8QCc/mqdefault.jpg)
Thank you. I’ve long wondered how blended shadows were created. It’s one of those subtle effects that really drive the realism to the next level.
Dude this is literally genius you're amazing! I always had hacky ways of dealing with double shadows and have never seen such a good method for dealing with this. Thank you so much
Really appreciate this tutorial. I found some cool tips for integration. Thanks a lot for your time!
I usually work with blender + After effects and really want to learn nuke, i love the tracking tools in nuke, interested to see more 🙂
I almost emailed you 2 days ago asking if you had any ideas about avoiding double shadows! it's like you read my mind.
Intelligent methods sir keep it up
great tutorial dude! this method of dealing with double shadows is awesome!!
Dang Bro! amazing stuff here, this doubke shadows can be a huge nightmare to comp!
Thank you so much for this.
Great Knowledge.
Great tutorial
Awesome video! The technique with creating a shadow cleanplate in the hsv colorspace is really interesting!
Thank you so much. Your tutorials are very good. Please could you make a tutorial on the noise node and it's goodness, cause I'm trying to understand that node very well. I know it's very powerful
Thank you! Yes I plan on doing more creative oriented tutorials in the future which will include noise!
Thanks for this valuable knowledge 😁
My pleasure!
Looks awesome. Thanks.
for crossing shadows I use a constant/voronoi gradient to fake the real one and min this with the cg mask as mask input.
The rest of implementation is neat though
Interesting! Would be awesome to see a demo of that!
@@CompositingAcademy will do a rough sketchup... brb in some mins
pasteboard.co/K3yn05n.png
pasteboard.co/K3ynCgc.png
The way how you create your shadow in this complex case is really great; "min"-ing a shadow can help in a lot of simpler cases to add the correct shadow in a quick and efficient way, though and definitely without bothering "stencil"-ing the shadows or paint them away
Thank you so much
Sorry if i'm asking something after too much time. I'm working on a shot, and I'm trying to create a "shadow cleanplate" similare to yours, but when I apply the keyer effect directly to the colorspace node it doesn't work as expected, it just make darker the dark areas. I get something similar deactivating the red and green colours on the node
Ohh this is awesome, thanks for sharing!
BTW I saw you needed to use Idistort before Itransform, right? Is that needed to transform to work properly?
Great Video
Excellent ❤️
Nice tutorial, I am kind of confused about the part where you set the colorspace to HSV but then used a regular luminance keyer to key the highlights in HSV.
Should not the keyer be set to blue keyer, or am I missing something?
You can set it to the blue keyer, that would be correct. In my script here, I copied back in the Red and Green from the original, which is doing the same thing but in a different way.
Hey Alexander, Thanks for you tutorial. Very interesting way of adding shadows.
This reminded me of a tutorial from FXPHD Tips and Tricks Volume 3, class 05.
This is also about adding shadows and this technique involves the exposure node.
I would love to know your take on that. Probably your way matches the attenuation of the shadows more.
Thanks! I haven't seen that particular tutorial so I can't really compare. But it sounds like they're just using an exposure node instead of multiply which basically do the same thing, except that the exposure node is measured in camera values ("stops" of light), vs just 0-1 like a normal grade. Visually though there's no difference.
@@CompositingAcademy thanks! good to know
Thank you very much. This Flatten shadow technic was explain in a blog somewhere but could not replicate it correctly. Now I can. Thx again.
Perfeit teacher, my heart for you! Lembro de mim? kk
Can i do this in non commercial version
yep you can! The free version has very few limitations
Thank you very much for posting this- great information here!
I was curious on another topic. If you're using a Nuke spotlight in a 3D scene, how do you get the spotlight to create soft cast shadows like you'd see in nature? So far all I've been able to get are hard shadows with the spot or directional light. Thanks in advance!
Unfortunately there's no easy way to do a soft shadow in Nuke as far as I'm aware. There's iBlur which can simulate a soft shadow easily, but if you wanted to just cast soft shadows you would probably need to build a gizmo that does this. I have a few ideas on how you could probably build that tool, but it would be hard to explain on a text.
Hopefully though they just add some more features to Rayrender so we can get soft shadows in a less confusing way.
thanks for sharing :)
btw do you have a planning of your tutorials? I would love to see your take on colour matching Footage with CG elements
I have a few in mind, I'm currently making some "bigger tutorials", but these take quite a lot of time to produce.
I PUT A LIGHT IN 3D SCENE AND THE OBJECT DONT PROJECT SHADOW ON THE PLANE WHY?CAN YOU HELP ME?
Cool thx!
This is a neat trick. I am staring at that HSV image though and I'm noticing the shadow is purple but the floor is blue... Isn't there a From merge node trick in there somewhere that could be used with that colourspace? i.e. painting the tree shadow to blue first and using the difference info somehow?
You definitely could. What you're seeing primarily is difference in hue which is why it's appearing more red. Basically the warmer areas / areas in the light are warmer in hue, so you could do similar to what you said or just do some color grades which might be faster.
HO I CAN DO SOFT SHADOW FROM THE LIGHT IN NUKE?
Could you please tell me why did you reduce Gain to 0.5 with luma key, why not 0 ?. Is there any advantage of doing more steps with less reducton ?
You probably could do a lower number and do more “steps”, personally I haven’t tested it. It would mean a lot of nodes, but generally 0.5 and just copying hasn’t produced any obvious color banding for me in most cases.
Worth testing out though, might give a good result!
Asking the real question, where did you learn compositing? Just curious to know 😊
I’ve been doing vfx since I was young, but also had few great instructors in college (full sail). The primary instructor I had no longer works there.
However a lot of the advanced stuff I learned mostly from working with other people in vfx studios, as well as with VFX supervisors at places like ILM or Weta where they have 20+ years of experience. Some of those guys have incredible artistic eyes and technical understanding, so when you spend a lot of time with them you learn exponentially
@@CompositingAcademy Holy cow you worked at ILM and Weta. That is a very proud moment Sir. As a person living in third world country, I can't study VFX because of no VFX college. That's why UA-cam and your courses helped me out. But I don't know how would I learn these things from scratch or land a job. Thanks for replying me.
Awesome I’m glad they have been useful! My goal is to make a competitor to vfx colleges because they’re vastly overpriced and many students leave with an insane amount of debt.
I believe I can put something together that will be better than those programs at a fraction of the cost, so I’m working hard on that this year! The courses were structured in order for that reason, but I will be adding some additional elements to finalize the whole package as well as having live teaching included for those who want it.
@@CompositingAcademy The debt part is pretty scary although I agreed that learning VFX is pretty expensive. You are doing pretty good with the quality of NUKE series specially NUKE 707 has Transformers like CGI quality. I think you are not just a compositer but also a great CG Generalist because I remember you modeled porsche in NUKE 303 which was fantastic.
The effort you are putting in NUKE series are astonishing. I have a small question and an obstacle which is bugging me alot. Like compositer role is to get all final results from other artists in order to complete his work but how would I make a project or showreel because I can't add the assignments I did from courses in my reel. I hope you are getting my point because how would I tackle other roles in it or should I collab with other students from different field for making a reel. Thanks in advance.
I really want to learn your tutorial, but my English is not good. Do you provide multilingual subtitles?
Hey Edmund,
The Udemy beginner courses have english subtitles, but not multi-language. However, if you google for Udemy subtitle translate, there are ways to do it.
Cheers!
@compositingacademy can u make it more begginer way please i really need this
Sorry i'm not sure of a way to do this easier, this is more of an advanced tutorial.
Awesome! How this can be done in After Effects though?! I'd love to be able to achive this in AE, any tips? Thank you in advance!
The main way would be figuring out how to get a good shadow cleanplate, and then just blending that into the shadow alpha. You can roll off the highlights in a similar way probably by using the Curves tool in After Effects, and just repeating that effect until you have a similar 'flat' image. Would have to play around with it as I haven't used after effects in a few years, but the principle is pretty much exactly the same: target the highlights, bring them down a bit, and keep doing that over and over to get a shadow plate.
@@CompositingAcademyThank you so much for your answer! I'll try it!
Dude it's just a cube 🙄.
Btw great video! Keep it up! 😁
POV COMP SUPERVISOR : 😒
Cool
Very good . Is there any particular reason to use HSV intead of HSL in the colorspace transformation step? great content, thanks for sharing that
Thanks! I think HSV and HSL are pretty much the same, the math is slightly different. Ben Mcewan wrote on his blog that explains the difference pretty well:
"HSV is the same as HSL, except the V stands for “Value”. There is a difference between luminance and value based off the math used, but it’s a similar concept. If you multiply the blue channel in HSL vs HSV colourspace (blue channel, because Blue has been converted to Value), you’ll notice that HSV produces a similar result to just multiplying your image the regular way, whereas HSL appears to desaturate the highlights as they get brighter. HSV also seems to give a flatter, more diffuse look to the saturation values, so you’d probably want to use HSV over HSL in most cases."
@@CompositingAcademy thanks for the reply. I often use HSL. Going start to use HSV more times now. Besides that, one thing came to my mind: this "isolation process" (let's call this way) by changing the colour space then changing back to the original, shouldn't isolate hue saturation and Luma (or value) information in each isolate channel R,G and B? so why this difference between HSV and HSL affects the saturation since the channel for that is the green one? does it make sense?
@@elironerosa3609 Hey! Sorry didn't see this comment before. I'm not entirely sure to be honest, I think the way the "Luminance" channel is created must be slightly different, which is why it is affecting the saturation slightly when using that alternative option. My guess is that it's doing some math behind the scenes on the R G B channels, but these two colorspaces do some slightly different 'difference' operations to create that brightness channel. In other words, it might be "weighting" one channel more than the other when it goes into the highlights, when calculating what is the luminance vs color. Probably someone who knows further in color science or the guys who work at foundry could answer better than that, but that's about what I know!
I sidnt under stand clean plate part plz hekp
Is it s spike fission -> fusion?
Has anyone been able to get this tactic to work? On my end it doesn't really do much to highlights, and just makes the image WAY darker without revealing/leaving the details behind. Exact same node setup. When googling, looks like another person had same issue.... working in linear etc.... !
hmm usually it works, if your highlights are completely clamped and there's no more detail in them - then there's no way to bring in more detail without painting the detail in. If your footage has some dynamic range it should work. If there's detail in the brighter areas like the example image ground area, it should also work.
Wondering, how did you get the shadow from the tree to spread on the box though?
that was rendered in lighting. I had a directional light for the sun, then I just used a geometry to block the sunlight where the tree is placed, to cast a shadow.
Tricky to get the angle lined up with the real shadows but it works
@@CompositingAcademy Ok then what did you do about the original tree shadows on the floor? Roto them out and replace them with the 3D shadow?
@@belovedangels6457 I believe I used light linking / shadow linking. So that the tree shadow would only cast onto the cube but not the ground as to not create a duplicate shadow. Then the cube would still cast a shadow. Been a while since I did this tutorial, but if it's not that then the other way would be to just render the shadow pass separate (disable the trees and only render the box shadow).
treasure
The project file is broken for download. It gives a screenshot for download
It’s a download page, you just type “0” in the box and hit enter, and it will give the project files
It takes a 3MB file to download@@CompositingAcademy
the backdrop is a still image here, does this method not make the background plate very noisey.
This method works on video as well. You would just framehold the video, create the shadow cleanplate from the still image (the same as explained here), and project that out onto 3d ground geometry.
@@CompositingAcademy ah that makes sense, amazing thankyou
@@scatmancrothers no problem! cheers
Can i do this on other applications? I only have Blender
Hey! Most definitely you could, in After Effects or Fusion probably would be the easiest. I'm not too familiar with Blender's compositor, but it probably could do this. Honestly though it's best to learn either Nuke or Fusion for more advanced compositing. Nuke has a non commercial version that is free to use for personal projects!
@@CompositingAcademy thanks dude!
👍👍👍👍
Hello i am big fan please make new courses nuke 505 for deep compositing in nuke i am waiting and my friend's also waiting your course
Thank you! More stuff coming soon I promise!
@@CompositingAcademy i am Very excited
just fraking show the "good way" and thats all... Jesus 5 mins in and nothing happening yet