How to remove vignettes from 35mm footage

[[gk]]

Using manual aproaches such as post applications vignette tools and what not, includes the element of eyeballing thus inacuracy.
Here is a way to remove that element of sloppyness.

html http://www.droschler.dk/Tut-Vignette_removal.html

Enjoy o/

[PaPa]

Great tutorial. Thi would work well under certain conditions and not in others. For example, by adding additional exposure , you also increase noise, lower contrast ratio and the corners will have a very different look than the rest of your image which may or may not be distracting.

But still a great way to remove some unwanted vignetting depending on how extreme it is. I imagine this would work much better for slight vignetting rather than hard vignetting.

[[gk]]

As long as the pixel data in the original frame isnt clamped, this method works for all situations.

All I do is raise Brightness to get to 255,255,255 be that in legacy mode ofcourse! - in the hotspot area, thus phasing out the vignettes in a proper blend.

You are right about the contrast loss over frame due hard vignette boost. But I dont think people have THAT bad vignettes going

Anyways you can add 2 masks to counter the loss with the control isolation vignette, be that a masked exposure up and masked contrast up to gain a full uniformity.
As we got the isolated Vignette we can use it to control any CC via mask.

Now how can we control that the masked vignette recives just enough exposure/contrast?
Simply by adding 2 float nodes to two additive masks and set a global difference which should output 0 everywhere there is intergity. then all it takes is to poke the contrast/exposure at float till we got a 0 output


I dont have that in this example however.


Infact I will add that to proff a solid case.
thanks for addressing the missing aditional masks!

[Will Clegg]

If I understand this correctly, in order to accurately use this method, you need to shoot footage of the white board while the camera/35mm adapter is still built. Otherwise, if the rig has been taken apart and then rebuilt since the original shoot, there will likely be some variation in the vignettes created.

I think it is implied in the way you wrote the tutorial, but I missed it the first time - you might want to highlight that the recording of the reference frames needs to happen on set, and will not be useful if recorded days (or weeks) later.

Otherwise this is an excellent tutorial for anyone suffering from the dreaded vignette. Thanks, Mads.

[[gk]]

Cheers Will!

Im poking on highlighting key elements such as the one you mention. Which also includes grabbing new isolated vignettes on new f.stops/lens swaps even.

Poking on the node masks to get a NULL difference thus upping contrast in the bightened areas as well.

[Chris Santucci]

Nice, but this tutorial hurts my head.

Couldn't you just export frames from a sequence, open one in Photoshop, record an "action" while applying a suitable dodge, then run the action on all the frames and re-import into the NLE?

.

[Will Clegg]

There are many ways to do this by eye, including working outside of an oval vignette in a secondary in Apple's Color or other CC software.

Mads's tutorial uses actual data from the specific vignette created in-camera though - which will make a more accurate correction, although a bit more complicated to implement.

[[gk]]

Correct.

Im not aware of the amount of After Effects users here, but I aim to write a custom effect setup you can load on the footage.

It doesnt take a long time to do manually as soon as you know what you need to do, I belive the additional small extra work is worth to ensure a close to mathematical correct phasing of the vignettes is worth it.

I for one is a post person/cg as well and I spend long times painting textures on mesh that will only breefly apear on screen, so my tollerance towards whats relevant and whats not relevant might be a bit more variated than other peoples tollerances/acceptance.

I perfectly respect if people want to do it by eye, but also note that its inaccurate.


Regarding a recored macro in photoshop..hmmz.
I only see a problem where you would need to have a variable depending on the base peak hotspot average pixel value
In the case I worked on in the example I belive the shot had a desaturated peak value of 133,133,133
This means I needed 2 legacy brightness in serial as a brightness increase only goes as far as 100 tones, thus 133+100=233 where we need to hit 255
So an additinal brightness legacy raise of 22 is needed.
This would be the variable in the marco script I wouldnt know offhead how to solve automaticly.

A general "lookup" on pixel values could automaticly estimate the center of hotspot and then subtract that from 255 in my case 255 ( read 133) average =122.
Then aply that automaticly in Legacy mode.
Viola done.


About hurting heads! I will look at it and reformat it for a less disturbing read session.


I also think it would be wise to add a general node based big visual drawing of the process, so it gives meaning and would be fittingbaleishblegh! in any software.

[Apefos Adapter]

check this thread to see a easy way to correct vignetting

http://www.dvxuser.com/V6/showthread.php?t=144255

[Michael Y Wong]

Thanks for posting this.

Very insightful, we are still using the quick and dirty fix of simply 'zooming in'. Your approach has definately got me thinking.

-MYW.