Tag Archives: The Foundry

Blender VFX wish list features


EDIT : the project started at http://ocl.atmind.nl/doku.php?id=design:proposal:compositor-redesign . Please donate and help the project

I have been at the Blender Conference this year & had the chance to met a lot of great guys as also talking with Ton about doing vfx with blender. I attempt the “VFX roundtable” talk… yes, I was the annoying guy who stoll the microphone the entire time 🙂, and yet didn’t have time to give my entire list there. So after talking about it with Ton, he asked me to send him an email of it, which I did.

I tried to make it as much detailed as I could and also to not making it too personal as well but keep it as a general need basic features.

AndrewSebastian & Jonathan asked me to post that mail on my blog to share it with the community, so here it comes…


When opening this kind of topic I know that a lot of Blender User like to throw stones at the guy writing it :), so here are some facts before starting. I’m working on a every day basics with After Effects and sometimes I do use Nuke. I’m a TD Artist for a video game company working on off-line & realtime production, so my job is to design pipeline and workflow or either new tools to make a production more efficient and comfortable for the artist (means I’m use to work with developers a lot, and know how much a pain in the ass I can be sometimes… trust me 🙂 ). On the side I’m also doing Matchmoving for a living also as VFX compositing & some Color Grading too. And those days I’m also developing  scripts & filters for After Effects which are distributed over at AEScripts.com (mostly stuff to enhanced workflow -sometimes inspired by Nuke I must say :p-).

All this doesn’t mean that I know stuff better than someone else at ALL !! I just try or I’d like to try to use Blender as much as I can in my Freelance time, and sometimes I just can’t because basics features are missing, and that’s what I’m going to point out here (or at least try).

I do understand that we don’t want to make a copy of AE or Nuke in Blender and I do agree that this would be totally wrong. But since there are leaders in this field and I know them pretty well, I will mention some of there features or workflow as illustrations & ideas. So you will have to take my words as “something like that…” rather than “it has to look like that!”. Note that most of the features I’m going to talk about are present in all compositing software not matter if it is nodal or layer based.

So you wonna do VFX with Blender ?

One thought we had with Sebastian at the Blender Conference, which I believe illustrate all those wishes pretty well, is “Try to do a VFX tutorial for Nuke or AE using Blender, and if you can’t do it, it means you are missing the tool”.

A great website for this is VideoCopilot.net ! And I’m NOT talking about the ones with third party plugins but actually the one that does use the basics provided tools. (I’ll mention some of them later on).

VideoCopilot’s tutorial are great in way that they are very professional looking, accessible to anyone in the field, and covers what you would do on an every day basis as a VFX Artist. And usually you would be able to do those tutorial with any Compositing software as far as you understand the logic behind it. Tools looks sometimes different, but they are all there in all the programs (except Blender for now, and that’s what I’d like to change).

I don’t need evidence that Blender could do the same by doing some tricks and finding some way around the problems. I’m always looking at an efficient, production ready approach of the problem. For instance, recently the RotoBezier Addon came out, and while this is a great and smart trick, to me, it is just a work around in the mean time, but not at all production ready for several reasons that I could mention if you asked me :).

Ok so now let dive into the list !

VFX Feature list for Blender

Color Management

As for today the CM (color management) is not working very well and too hidden IMO.

I did talk about it with Matt Ebb a lot. Even if we have different opinions about it, I know he has good ideas and great libraries to use in mind, but not so much time to work on.

His philosophy is to have it completely hidden in background and doing all the work for you, which I deeply disagree with (but again just my opinion). CM is good for having LW (linear workflow), and LW  is important, but CM is also important for Color profile and Color models and you really have to know in which space you are currently viewing stuff (linear, sRGB, … ).

Today I can never tell if my viewer is showing me in linear or sRGB, and same for color picker.

In Nuke you always have an option in inputs, outputs or scopes to choose which color space you want to look at (ie screenshot) so you can have a complete control over it.

So having it hidden and automatic is ok in case I can overwrite it anytime I want to !

color profile selection in Nuke's reader (input)color profile selection for Nuke's viewer
color profile selection for Nuke's viewer

And as I mention at the roundtable, you don’t always need linear workflow (and thats where Matt is going to disagree with me 🙂 ), for instance it is usually more a pain in the ass for Color Grading than anything else. And I have been talking with some professional colorist who told me there big color grading workstation are working in non-linear (ie Davinci Resolve). Which don’t give me wrong can work with linear too, but if you take something like the Color Balance Node which Matt and I develop, the formula I used works better in non-linear. Same formula used in Magic Bullet Colorista and very close to what use Color grading suite by the way. So LW, great for compositing shots and playing with Render passes, for video edition & final color grading, not so much.

EXR Full Support

Support for reading any custom passes :

At this point Blender cannot read any passes that it doesn’t support or create itself. So when I’m working with external client who works with other pieces of software I cannot read their passes (except the normal kind, diffuse, spec…)

In any Nodal based compositor I know, there is a special node called “Shuffle” (or whatever) which separate the passes. The reader (input) just gives the default pass (beauty ?), but the shuffle does the job of selecting a specific pass.

Shuffle node in Nuke

Support for writing custom passes :

Now I have to create a new RenderLayer with a material overall and save it to another file. when actually what I’d like to do, is create a new pass, name it and affect a material overall to this pass. all that in my default render layer. (this part is not just EXR but also rendering related).

Support for reading/writting attribute :

You can read or write attribute in exr. Some kind of metadatas which can be very useful even to pass datas to your compositor as for instance the actual position of the camera or values of the distortion, anything you can imagine of that can be useful (not as important as the two above, but again full support is better 🙂 ).

Expression Py

I am so missing that !!!
Basically it means that for each params you have you can set a script. In AE it is called “Expression” as in Nuke by the way.
The idea is to right click on a params and then select “set expression”. Expression can link to everything in blender other params, maths, … so at the end you can do some really crazy stuff as constrain or operation or conditions…
We are using that ALL THE TIME
In Nuke you also have an expression node which can gets several inputs and then you can write python expression by channels. It becomes almost an image processing or kind of pixel processing kind of node. Very very powerful :).

Nuke's expression node looks like that, but creating expression to any inputs like in AE sounds more powerful to me !

Enhanced Group

Groups are actually not that bad in Blender, but a bit too rigid !

  • One thing I’d like possible is to choose the inputs I will have in my final Collapse group UI. Today to do that it is a bit hard and not very flexible. Plus you cannot add special UI inputs which might not be a node but just a value.
  • I’d like to keep some group expand but still being able to work with nodes which are not in this group. Right now if a group is collapse you can only work on this one.

A nice recent approach on this topic : http://www.youtube.com/watch?v=OUidTgzy8zo

SDK, Py, OpenFX

I did develop the Color Balance node in compositor with Matt. Actually I must say Matt did most of the job just because I couldn’t understand the way making UI node was done in C. First because I suck in C, secondly because there were so many file to change I was totally lost (register, type, processing, name, RNA, DNA, …. all that different files). So while I understood pretty well in C the part where the pixels got process, all the UI was a nightmare for me, and Matt did handle it.

I can only imagine how many people get stuck by that while it can be very easy to do some pixel processing just by typing a few lines of code. Several solutions for that :

  1. A well documented SDK (or “BDK” ^^) which as a bad coder I usually hate, too C related for me, but good for developers who wants to make plugin
  2. Python script/expression. As mention above being able to control a input via a python expression or even better an “Expression” as mention above as well
  3. Being able to load/write GLSL code for processing ? I discovered Pixel Bender for After Effect, its kind of a glsl-like for processing image in AE, VERY easy to code, VERY powerful with possibility to add user UI inputs in the shader which AE will transform as interface ! This is perfect for a crappy coder like me and so great in “production” environment. you can prototype things or solve problem in no time.
  4. OpenFX, this is my personal favorite. I have a full thread about it on my blog ! Ramen did implement it, it even did port some Blender’s nodes to OFX already. For a developer stand point of view it means he can create a plugin for blender, and make sure it will also work on any other software that does support OFX (Nuke, shake, …).

But more importantly for somebody like me, it means I can buy a plugin like Keylight (for green screen keying, because lets be honest Blender’s keying tool are a bit old) and use it directly with Blender, without the needs to buy After Effects or Nuke to use it. PLEASE read my blog post about it which clarify the point on that : http://www.francois-tarlier.com/blog/index.php/2010/01/openfx-ofx-an-open-plug-in-api-for-2d-visual-effects/

-> I’m doing some filters for After Effects (http://aescripts.com/category/plugins/francois-tarlier-plugins/) , and mostly because I cannot do it with Blender, it’s way too complicated to achieve the same thing !!

UI Design

Ok I plead guilty on this one, it is more of a personal wishes that something else or VFX related. At least I get a point for honesty ?

  • Straight lines !!! Like Pablo called them the “spaghetti lines”. This is so rigid and messy at the same time. It makes you work only horizontally and from left to right, you cannot redirect lines like you want. I hate it ! But maybe it is just me ! Take a look at one of my comp here : http://www.vimeo.com/11279314 ! (yeah I know, If I had “expression” I wouldn’t have so many nodes but… hey… you got my point 🙂 ) As I recall, Digital Fusion does straight lines with angles by itself, I hate it too. Nuke (ie screenshot) does straight lines left to right, right to left, top to bottom, so on … which you can break by pressing CTRL and adding a “dot” on itThat is what I call flexibility !!!
  • Plug lines into dot socket is a bit old and rigid too. In the below example (same in Fusion), each node would have an arrow for each inputs it can get (w/ name on the arrow). Makes it more flexible to layout (it goes with the above point)
  • Do not put controls in the nodes, keep it in the panel (float or dockable panel). Blender kind of have both now, but I’d like the option that if I double click on a node the panel would show up for instance ?

Break the wall between compositor & 3D view

Today every compositor as some simple 3d capability. Blender gets both features, but it is impossible to use it together. Why would you need 3d in compositing ? (best reason here : http://www.youtube.com/watch?v=upG81s75UD4#t=7m13)

Let say I have this render of a street w/ 3d and matte painting stuff. And I want to put some guy shoot against green screen in this street, this is what I would have to do in Blender to get the shot done :

  • import green screen footage in compositor
  • keying it
  • render it
  • place 3d plane into my scene
  • add a material
  • create a texture with the rendered keyed greenscreen footage
  • apply with UV stuff
  • probably render those plane on seperate passes/layer than the enivronement in case I want to do some more fx/compositing on top of it.
  • import that result back in compositor
  • do all the compositing magic

Do you see the problem here ? so many steps, so many in-between render… when actually all I wanted is pretty much place those keyed plane/billboard in 3d space with a bit of orientation and getting the camera data from your 3d scene !! w/o in-between rendering !

Also know we can see more and more of what we call “Deep image compositing” which basically use a point pass (or position pass) and generate a 3d point cloud based on this pass. it helps for all kind of things (preserving edges, getting geometry information, occluding, relighting, …. )

(ie http://www.youtube.com/watch?v=upG81s75UD4 ||  http://en.wikipedia.org/wiki/Deep_image_compositing || http://www.fxguide.com/fxguidetv/fxguidetv_095/)

Real Camera

lenses, camera sensor size, distortion, … all the good stuff 🙂 (Matt already started some work on that http://mke3.net/weblog/blender-sensor-sizes)

Custom Passes

As I mention earlier, being able to create custom passes which we can assign Material. Pretty much like material overwrite function but into passes. And also with some additional passes by default as Point pass (world and object), and normal pass (world, object & camera).

Masks (roto)

I still can’t even imagine how we can do compositing without masks. If there is one feature above all I’m ALWAYS using in ALL my comps its masks. And to be honest not having it, is one of the main reason I’m not using Blender as primary compositing tool today.

I know someone did talk about some fancy/geeky/gadget mask tool he would love to have in Blender instead of simple/basic vector shape with handle. My opinion on that ? Keep it basic at first ! Simple tools are always good to have first. And even if it might be tedious to do, at least we know it works in all cases.
In case you really want to go with the geeky automatic paint / rotobrush tool, take a look at GraphCut & SnapCut algorithm. That’s a paper from siggraph which is used in AE now know as “roto brush”. Which is a great tool by the way, but from my experience, not yet better than a good old vector shape 🙂

Image Cache / Dynamic Proxy / Pre-cache preview

VERY IMPORTANT !!! you’ll never be able to do some good VFX compositing w/o it.

DJV does that pretty well actually. you can render the sequence into RAM to play the preview in realtime (Again I still don’t understand how we can survive w/o it)

Downgrade the viewer (full, half,quater,.. resolution)  to save some memory or put more frame into RAM would be great too.

Press a button > render to RAM > play it !

Interact with Elements in the viewer

Being able to select one of the “layers/node” and rotate it, scale it, move it,…. directly in the viewer. (like you would do in photoshop for instance (yes yes it is possible with node base too of course 🙂 )

2D Tracker

I’m a matchmover, so trust me I know how cool it could be to have a camera tracker in Blender. I’m following the libmv project very closely and talk a lot with Pierre Moulon, one of the active developer on the project (since Keir is kind of busy those time).

But the fact is I don’t really need a 3d camera solver. First it would take a long time to have something which works great and be production ready. As for today I do have syntheyes, which is not expensive, and that I can use.

BUT, having a simple 2D tracker in the compositor to do things like that for instance : http://videocopilot.net/tutorial/set_extensions/ that would be PERFECT !

I did talk about it to the LibMV team and they would be glad to help, but they would need somebody experience with Blender C code to implement it quickly. They asked me for a proposal a year ago which is here :




It’s not everything, but it is quite a lot for now. Some of those request are more important than other, and I would be glad to discuss this topic with anybody.

One tutorial that does sum up all those features is this one http://www.videocopilot.net/tutorial/blast_wave. Bring the missing tools which will let you do that, and you’ll be much closer to be a “production ready” VFX tool… IMO :).

I would also encourage you to take a look at Nuke PLE (personal learning edition), it’s free and complete (only rendering are watermark.
Also take a look at the work done on Ramen VFX, Estaban did some great things there
And finally, make FxGuide.com your favorite website and watch a lot of breakdow 🙂

Nice VFX Breakdown to watch :

[Bonus] Hair Simulation

This is not directly VFX related but as I understand, you had trouble with hair simulation on Sintel. I did talk about it with Daniel Genrich at the time, because I think he is the dynamics guru who could nailed this kind of thing. So I gave him a few papers to look at which he thoughts were very interesting and not very hard to implement, but as I recall it was too late to implement at that time and you guys didn’t have enough time to do it or even take the risk to test it. But I guess it is not too late for next projects.

When I was working on face tracking/recognition at Ubisoft, I met someone (Basile Audoly) working on mathematically modeling the dynamics of hair. He gave us a demo which was very impressive. And if I’m not mistaking, he has been working for l’oréal research after that 🙂

His program was able to model curly, straight, dry, wet, growing,… hair, but also dynamic cutting, all those features model in his research. I have personally seen his demo, and I have never seen something so real, see by yourself with the video !

Here are the papers and videos :

Super-Helices for Predicting the Dynamics of Natural Hair

Pixel Bender : ft-SSIBL (Screen Space Image Base Lighting) shader for After Effects


ft-SSIBL for “Screen Space Image Base Lighting” is based on a topic I covered in a previous post about Roy Stelzer’s “2.5D Relighting inside of Nuke”.  In this shader I tried to reproduced a few approach found in his Nuke script. So with a Normal pass (object or world), you will be able to do some relighting with a HDR map. The shader won’t compute the 9 coefficients (spherical harmonics) needed for you as describe in this paper : http://graphics.stanford.edu/papers/envmap.

The default value are from the Grace Cathedral, San Francisco Lightprobe http://www.debevec.org/Probes/

Download & Sources

I will for sure add more feature to this one in the future, so come back check it out

How to install it ?

Just copy the .pbk file into your “Support Files” folder in your AE install directory

How to use it ?

How to convert a Light probe into the 9 spherical harmonics coefficients ?

Here is the tutorial


2.5D relighting in compositing with Blender (using spherical harmonics)


About a year ago, I was watching a great presentation at the Nuke Master Class by Roy Stelzer (2D TD at Double Negative) about “2.5D Relighting inside of Nuke”. There is 3 videos which you can download from The Foundry and also files which goes with. I recommend you to watch those videos especially if you are not familiar with relighting in post-production.

What Roy did, is using a paper from Siggraph 2001 called “An Efficient Representation For Irradiance Environment Maps“.

“An Efficient Representation For Irradiance Environment Maps”

This paper cover how to do Environment Lighting based on Angular Map (or light probes, or HDRI map, whatever you want to call it) but without using those map for computation. Odd you said ? Yes sure !
They develop a technique using Spherical Harmonics to simplify an HDR Angular Map to only 9 coefficients !! That means after their process, you only have to do a few multiplication with those 9 coefficients and then you are done !  Check out the comparison  :

left is a standard rendering, right is using the 9 parameters. Grace Cathedral Light Probe was used for this rendering http://www.debevec.org/Probes/

If I’m not mistaking, there is less than 1% of error between both rendering, and of course you can image that the second one goes much much faster 🙂

Please visit there website and presentation for more information about it :

What did Roy ?

In his demo, Roy shows the use of  the irradiance only of a Angular Map. This means he will get all the HDR Luma intensity of the map (no color variation). Once you have the 9 coefficient, the operation is pretty easy and so it compute very fast. The operation looks something like this :

color = c1 * L22 * (x2-y2) + c3 * L20 * z2 + c4 * L00 – c5*L20 + 2 * c1 * (L2_2 * xy + L21 * xz + L2_1 * yz) + 2 * c2 * (L11 * x + L1_1 * y + L10 * z)

Where c1, c2, c3, c4 and c5 are 5 constant number given in the paper. X, Y, Z are float numbers from your normal pass (X2 means x*x ; xy means x*y ; so on … ). All the L22, L20, …. variables are the 9 coefficients)

So as you can see, this is not a really complicated operation and it does compute really fast. Running this on each pixels would return a kind of light map (irradiance) which you can use to ADD to your original color map.

What did I do ?

As I did for the Lift/Gamma/Gain the first time, I tried to reproduce this formula with the “Math Node” of Blender. So for that matter I did use the EXR file (render passes) provided by Roy in his demo, and only kept the Normal and AO passes.

Blend File here

This is not as fast as it could be, the render time for a 1920×1080 is around 1.5 second (well for HDR environment lighting we have seen worst ^^). There is several reasons for this to be slow, but I’ll come to that later.
Note that for this example, I did use the Grace Cathedral Light Probe value and not Roy’s light probe.

I was kind of happy of the rendering though, but a bit disappointed to only get Luma value when environment maps have so much information about colors as well ! (you thought the previous example was a mess with nodes, wait for the following one 😉 )
UPDATE : I was totally wrong 🙂 !!! the only reason why I only get Luma (or actually greyscale) is because I used Math node. I thought it was able to do the operation on any kind of input, but actually it does it on only one composant. So the vector operation never happen :p. I just figure it was possible by trying this same formula in another shader language (pixel bender) and see color happening ^^ . So my bad, Color works too, and I’m not sure to know the difference between the vectors or the Matrices in this case, except using the formula with vectors is much faster ! (I’ll change the blend file later)

So I took a closer eye to the paper, and especially to the example they provide, and I found out that their filter wasn’t only generated coefficients, but also Matrices !!! This means you can do the operation with 9 coefficients to just get the irradiance of the environment or do a similar (but a bit more complex) operation with 3 4×4 matrices (red,green,blue) !
I guess the obvious reason Roy didn’t go for this solution was because the computation is more slower, and he didn’t really need it since he is doing a kind of 3 points lighting in his example.

As I said the math are a bit different ! Here is the formula by using the 3 matrices :

n = worldNormal;
color.red = dot(n , matriceRed);
color.green = dot(n , matriceGreen);
color.blue = dot(n , matriceBlue);

Ok so while this might look more simpler on paper, remember the matrices are 4×4 and if a dot product is quite simple, it is not the costless operation too :). Here is what it looks like in Blender with “Math node” as well :

Blend File here
Again, due to the heavily use of Math node, I believe that 3 second is not too bad, but I’ll come back to that later. Also the node I use to rotate the normal pass, is using some math that might slow the render a bit too and not absolutely necessary

This shows that the technique is working pretty well, but probably not production ready as it is in Blender since we are missing a few thing to make this work smoothly.

What would we need in Blender for this to be more efficient ?

  • More “Input” nodes : this might be one of the reasons why the rendering is slow down a bit. Because actual “Input node” only works between 0-1, and the matrix number were between -n and n, I had to find a trick. For this I used the “Math node” for each number of a matrix. Setting it to “Add”, enter the value in the first input, and setting the second input to 0.0. So the output would be equal to the first input. I only did that because I couldn’t figure another way to have negative and positive values in Blender compositing in another way. But this also mean that the compositor tell blender to do an operation for each input, can’t say it’s much optimized 🙂
    • Input Value that goes between -n and n
    • Vector4 Input (right now you can only use Vector3, but you usually need to work with alpha)
    • Matrix Input (could even be useful to do some quick and dirty convolution)
  • Expression node : ok now, I’m dreaming for this one ! And this is IMO probably the main reason why this is so slow. I believe that each time I’m using a Math node, Blender does treat them individually. Which makes sense though, but it probably make a lot of exchange with inputs, outputs, inputs again, outputs again, ….
    I would believe that sending the entire operation at once to the CPU (or whatever) and getting it back at once would make things different and much faster (but I might be wrong on this one !?)
    Anyway the other reason for this node would be … well seriously, have you seen the mess with all the nodes ?!?
    So a simple field, even just interpreted by python would be great !!!!
  • Math with Vector : Maybe I did something wrong, but I couldn’t do a “dot product” between vectors, which is one of the reason why I have all those nodes. I’m doing the entire “dot product” by hand and this is heavy.
    I wish Math could be use with any kind of input. But again, maybe I’m doing something wrong here
  • Passes : we need more passes! For this example we need an Object Normal Pass rather than a World Normal Pass. Probably not to much to do though, the only problem I have with the Passes system today is that they are all hardcoded in Blender, which makes it complicated to create a custom one like you would have in Maya.
    I’d like to be able to assign a material overall to a define passes, but yet I would probably need to write shaders, which implicate the needs of writing shaders as well for the renderer. I guess OSL will fix that in a future if it gets implemented one day
  • Better support for EXR : beside this really annoying, flip node you have to add when working with other packages (I know flip node is nothing, but when working with 2K or 4K it is not the same deal, especially when the composite gets complex, you want to save any operation you could) , but I believe anyone is aware of this now, the other lack of EXR in Blender is the passes support. it doesn’t support custom passes coming from other package. Roy provided his Nuke script, with all the EXR file so you could play with it. But when I tried to load it in Blender, the input node couldn’t find all the passes inside, beside the usual color, Z (and maybe another one can’t remember exactly) it couldn’t find any. So I had to open the EXR in Djv, select the pass I wanted, and save it to another file as the RGB value. Really painful process

Hopefully someone hear me out there ^^

RollingShutter Plugin by the Foundry for After FX and Nuke Released !!!

I talked about the Foundry working on solving the rolling shutter issue on a previous post. It seams they were waiting the Siggraph to announce it. At the time I though it was only for Nuke, but the great news is they also port their plugin to After Effect CS3 and CS4.

It seems they fixed the horizontal panning issue, but still not the vertical one, which at this point looks impossible but I’m actually waiting for another paper on stabilization which should show up at siggraph too. Anyway it’s going to help a lot for matchmove !!!

Official RollingShutter by Foundry plugin page

“The Foundry” fixing the Rolling Shutter issue in some cases

We have already seen some Siggraph papers about fixing this issue in some cases (cf. horizontal distortion), but now we can see that The Foundry is working on it !!!

The Foundry is a great company known for their Nuke compositing software and also very known for their awesome “Keylight” plug-in as many others !

Check Out their blog and follow their research about all this stuff

Original Plate :

Fixed plate :