Doing things to a projected image

Hi Guys, I’m playing with projections and the new 3d tracker and want to know how to apply things like a an animated colour correction to the image being projected. I have a light source changing during the shot I’m cleaning up and want to apply a mastergrade to the projector but that didn’t work. How does one apply live image changes to a projected image

Hi Johnag,

Doing this in the live scene with the projector requires a little bit more work :slight_smile:

After projecting your clean image on to the receiving surface/geometry, you will need to grade at the composite level since this is a projection and not an actual diffuse map.

Bearing that in mind, I suggest the following:

  1. Complete the projection as normal for the clean up frame.
  2. Select the camera and add a Selective with a MasterGrade. (Apply to Camera and not surface affects the whole composite.
  3. If you grade now, the entire image will be affected.
  4. To limit the grade to receiving object, go to the Selective Controls and choose the 3D AOV controls.
  5. Looking at the Selective overlay by toggling F9, you can set your AOV to distance and narrow the selection to your object. This limits the grade to the selective based on the object’s distance to the camera.
  6. Any grades you make will be applied to the area of the projection in the context of the live projection and 3D track.

I have given far more detail than needed but this is achievable super quick in Action.

Hope this helps!

1 Like

WOW great tip Grant. Thanks.

One other thing related to this is that I added a gmask but I can’t get the softness to work for me with this technique. I can only see a hard line. I’ve switched the 3d shape transparence “for 3d only” and in the 3d geometry menu ran the transparency value to 100 but always get a hard edge.

What do you think I’m doing wrong?

Hi John,

I just tested 3D shape now.

I drew my 3D shape and added softness to the Gmask. I get a solid edge until I go to the 3D shape menu and change the Gmask Transparency to FOR 3D SHAPE ONLY. That works as expected with the softened edges of the 3D shape.

The things I can think of at this very moment is:

  • Check the Shape boundry settings. They should be SPLINE AND GRADIENT.
  • Do a Z-sort to ensure the render order is correct

If in doubt, create a new Action and just test the 3D shape and gmask softness to confirm it actually works for you on a simple level. Might help diagnose the problem too :relaxed:



OK yes all that works until you add a selective mastergrade to the camera as you suggested in the earlier post. Do you still get the soft gradient after you add the mastergrade to the camera?

Hi John,

I see what you mean. The 3D AOVs are doing exactly what you would expect. They are taking the object (receiving surface/geometry) and using their pixels as an exact point in 3D space. Understandably this is about Z-space and not pixel transparency.

So I decided to take another path and perhaps this is more successful.

You will still need to add a Selective MasterGrade to the Camera because you are affecting the result of the composite. However, in this case leave the 3D AOVs alone :relaxed:

Instead go to the MasterGrade node attached to the Selective and in the Shader menu, change the Selective option to MATTE instead of SELECTIVE. This will take the matte from Action and use that to apply the grade.

I tried this and it was able to keep the fall off-gradient of the 3D shape. To confirm the matte, go to the Outputs menu and look at the Matte Render pass in the viewport.

The only other way of achieving this with the soft edges is to output the matte from Action and use nodes in Batch to do the grade and recomposite.

Let me know if this helps!


WOW, thats a great insight but so clunky, no? I would never use that because its so counter intuitive in a batch type workflow. haha, sorry Grant i had to say it! choosing a object by its distance to camera and color correcting it makes my skin crawl! so fiddly. Its great theres an option i guess! i gotta have a negative review once in a while to get my stats up on logik, haha!

Hi Theo,

The one thing about Flame is that there are a thousand ways to “skin” the proverbial cat :slight_smile:

I was trying to work out a suggestion with John that everything could be done in one Action (and it can) but it just might not make the cut.

The internal method of doing it in Action (by any AOV) is different to what you may be used to hence it will make your skin crawl. But like I said, I was trying to figure out the challenge entirely in Action. I don’t have a preference as long as you can work successfully and Flame does the job!

What may suit you better is one of these two batch centric methods.

1- Colour correct the clean up frame before the Action node and use the context view to see the clean-up in the context of Action while you match the grade
2 - Colour correct the clean up frame after the action node by pumping out the result of the composite and matte of the geometry. You would use various nodes to grade and then recompose.

Just keep the conversation going and that will also raise your stats on Logik :crazy_face:


Thanks Grant, I got the results I wanted now by changing the shader selective.

This task was a simple clean up of an object on a wall and when the camera moved in the light level changed and so tweak of the levels going into the projector is all I needed.

thats one of things thats amazing with flame, the ability to change direction in a particular technique and attack it from different perspectives, i agree! theres no one correct way of doing it! It really helps in problem solving