Mask/blending for overlapping projectors?

1,134 views
Skip to first unread message

Stephen Harrison

unread,
Feb 3, 2022, 6:21:47 AM2/3/22
to QLab
I have a show where we are doubling up projectors to get more light. Due to lens limitations, the image sizes do not match, with one projector covering a smaller area than the large one.

This is fine, as we get more light on the central area, but it does give us a hard edge to the light from the second projector. Is there any way to blend/mask/vignette the image going to one projector only? I can't use the surface mask, as it's only one output that needs masking. And because the projectors are doubling up, rather than side by side, the edge blending doesn't enable.

Any cunning thoughts or suggestions?

Mike Post

unread,
Feb 3, 2022, 10:02:55 AM2/3/22
to QLab
Well, for an old school solution…. If you add some physical masking around the lens, it will create a soft edge “shutter” that might do what you need.  I’ve used Blackwrap, cardboard and sometimes just folded gaff tape.  Obviously you don’t want to put tape on the lens, but just barn door style masking around it could work.

Mike Post
--
Contact support anytime: sup...@figure53.com
Follow QLab on Twitter: https://twitter.com/QLabApp
User Group Code of Conduct: https://qlab.app/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/4b54e289-1e49-4d09-92f4-02c6cb97ea53n%40googlegroups.com.

Sam Kusnetz

unread,
Feb 3, 2022, 10:06:46 AM2/3/22
to ql...@googlegroups.com
Hi Stephen

Is the problem you’re seeing a problem of your image edges needing to be softened, or a problem of the raster edges needing to be softened?

Put another way, are you seeing the edges of the light beam of the projector and that’s what you want to soften? If so, that is not something that can be solved by any software in the world, since it’s a physical characteristic of the projector.

Physical masking is the way forward here. What I’ve done in the past with surprisingly good results is this:

  1. Tear a 6” wide strip of Duvetyne by hand (no scissors!) in the direction that it naturally tears straight. This leaves a rather fuzzy edge where you tore. Heavier duve works better, the heaviest you can manage to hand-tear.
  2. If there are any beefy threads that stand out from the torn edge, trim them away with scissors.
  3. With all the lights off in the room, but the projector on, drape the strip of fabric over the lens and look at the projected image.
  4. Align the strip so that it fuzzifies the edge of the raster to your liking.
  5. Apply large amounts of gaff tape.
  6. Repeat for the other three edges of the raster.

If your projector has a lens whose front element protrudes from the barrel, you might instead need to build a little frame for your Duvetyne mask because you don’t want to drape right against the glass.

Good luck!

Best
Sam

Sam Kusnetz (he/him) | Figure 53



--

Stephen Harrison

unread,
Feb 3, 2022, 12:16:31 PM2/3/22
to QLab
Hi Sam,

The issue I have is with the image edges of only 1 of the projectors.

The attached images shows the actual projected image areas on the screen. I can align the two images so that projector B shows the correct portion of the overall image, but there is a hard edge to the projector B image (doubled brightness). What I would ideally want to do is to apply a mask to the projector B output to get rid of the hard edge, and have a blended brighter section in the middle.

This is certainly technically feasible in software, but it doesn't seem like it's currently possible.

The hardware masking might be the best solution for now, although obviously not ideal!

overlapping projectors.png

Sam Kusnetz

unread,
Feb 3, 2022, 12:30:04 PM2/3/22
to ql...@googlegroups.com
Stephen

The situation you’re describing here has two hard-edge problems for Projector B, not just one problem.

The hard edge of the image, the picture or movie, is one.

The hard edge of the actual projected beam of light is another.

When you are projecting plain black, no cues, do you see a bright rectangle from Projector B in the middle of Projector A, with hard edges? If you do, that’s the situation I’m saying cannot be solved with software.

As to blending the image edge, that’s much trickier since what you’re doing is pretty unusual. QLab allows you to apply a mask to a surface, but not to an individual screen. What you’d need to do here is create two separate surfaces for the two projectors, then output the same cue to both surfaces.

If it were me, I’d crop down Projector A so that 100% of my useable area was covered by both projectors.

Best
Sam

Sam Kusnetz (he/him) | Figure 53

Stephen Harrison

unread,
Feb 3, 2022, 1:05:29 PM2/3/22
to QLab
There is no noticeable hard edge when projecting a black image on B. I realise that there is some light output, but it's not significant.

"As to blending the image edge, that’s much trickier since what you’re doing is pretty unusual. QLab allows you to apply a mask to a surface, but not to an individual screen."

Yes, exactly that. I'm well aware this is an unusual edge-case, and didn't think there'd be an obvious fix--just thought I'd mention it incase (and if I'm not the only weird one). Ideally, I'd have a wider lens for B and this wouldn't be an issue, but we deal with what we have! Also, cropping A is not an option as then it wouldn't be covering the area it needs to!

Best,
Stephen

micpool

unread,
Feb 3, 2022, 4:59:27 PM2/3/22
to QLab

This is fairly easy to do. (Apologies for the crap photography and the rough line up and projector mapping, but I literally threw this together in 10 minutes)

QLab.png

There are 3 surfaces:
 One Syphon output only which is where you send all your content to.
One for your Wide Projector
One for your  Narrow Projector set up as above
Mask in Mask well
Corner pinned to square and centre  with wide projector

A timeline group needs to be running all the time. This contains:
A camera cue with input from Syphon and outputting to the wide surface
A camera cue with input from Syphon and outputting to the narrow surface with scaling to match the image size to the wide projector (Bottom right pane above)

Video cues output to Syphon surface (Bottom Left Pane above) and you can crossfade etc.  as you would if you were outputting to a single projector.

The Mask looks like this
Mask.jpg

Without the mask the overlayed projectors look like this when protecting the grid
Grid No Mask.jpg

With the mask:

Grid Mask.jpg

When the content is projected the join is even less obvious

Sunrise.jpg

Doors.jpg

Obviously the line up is a bit off, but with half an hour or so to line up properly  you should be able to get it exact.


Mic

micpool

unread,
Feb 3, 2022, 5:06:49 PM2/3/22
to QLab
On Thursday, February 3, 2022 at 9:59:27 PM UTC micpool wrote:

This is fairly easy to do.

Demo Workspace attached 
Wide Narrow Projection.zip

Stephen Harrison

unread,
Feb 11, 2022, 8:25:02 AM2/11/22
to QLab
Hi Mic,

Thanks for the worked example, that makes a lot of sense and is certainly more flexible than what I ended up doing (gaffer tape masks on the projector), so I'll keep that in mind for next time.

I do wonder what sort of additional performance hit there is in looping the outputs back like this though? I was running multiple, overlapping ProRes 4444 clips with transparency, which my 2019 MBP was just about keeping up with. Syphon is suppose to be pretty low demand though, so maybe it would have been fine…

micpool

unread,
Feb 11, 2022, 9:38:49 AM2/11/22
to QLab
On Friday, February 11, 2022 at 1:25:02 PM UTC Stephen Harrison wrote:
I do wonder what sort of additional performance hit there is in looping the outputs back like this though? I was running multiple, overlapping ProRes 4444 clips with transparency, which my 2019 MBP was just about keeping up with. Syphon is suppose to be pretty low demand though, so maybe it would have been fine…

As always with these things it's a matter of testing your show set up with representative content, and then pushing it quite a bit further to know you have a reasonable performance safety margin, assessing the results by eye. A  test video with rotating bar and burnt in  timecode with frames is useful for assessing video performance visually.

You can attempt to quantify comparative tests using the fps logs in level 3 logs in the Console, or CPU in the activity monitor, but neither is likely to give a true meaningful result (Particularly CPU) ,  that is applicable to all systems.. If you try this with level 3 logs remember to turn logging off again after. 

I think it would be fair to say though that this order  of efficiency with a rough score out of 100, with 100 being the score achieved by a video playing straight to a single surface., is reasonably accurate.  (Starred scores are  for methods that would be applicable to what you want to do)

Single Video  direct to surface. SCORE 100 (for reference)
Single Video to 1 surface with 2 mirrored displays SCORE 92 (although this option would not allow you to mask one display)
Single Video looped back through Camera Cue to surface SCORE 85 (for reference)
Single Video looped back through camera cues to 2 surfaces SCORE 76* (Best Option for your applicaition)
2 Videos direct to 2 surfaces SCORE 56*

Where the syphon loopback really comes into its own is when you want  very large numbers of surfaces with the same content (or different areas of the same source content)

Single Video looped back to 16 camera cues to 16 surfaces. SCORE 45 (CPU Disk and Memory still coasting along)
16 videos to 16 surfaces SCORE 1 (Unusable, everything pretty much grinding to a halt.)

Mic

Sam Kusnetz

unread,
Feb 11, 2022, 9:47:46 AM2/11/22
to ql...@googlegroups.com
I do wonder what sort of additional performance hit there is in looping the outputs back like this though? I was running multiple, overlapping ProRes 4444 clips with transparency, which my 2019 MBP was just about keeping up with. Syphon is suppose to be pretty low demand though, so maybe it would have been fine…

Everything that Mic said about testing and safety margins is spot on, as usual, and I have only this to add:

Most of the processing overhead that Syphon causes is actually already being used internally inside QLab. When you tell QLab to output a surface via Syphon, you’re just making that internal Syphon process visible to the outside.

Just a little bit of info from behind the curtain.
Reply all
Reply to author
Forward
0 new messages