Led Panel Contrast

I’m currently experimenting with a custom led panel and can easily control it via DMX with vvvv (thanks for the help West ;-) )
Using a high contrast source like blocks of color show up really well but when I use a pippet and map the colors to the panel it lacks all contrast, mid colors appear too bright and mapping down the values doest do much apart from loosing resolution.

Does anyone know a way around this, maybe applying color curves to the source video, Increasing the contrast in vvvv?. (which i don’t know how to do in vvvv)

I’ll try and get a video up with some video being played back through the panel as soon as I sort out the lighting to record it well.
Heres what Ive got so far with some plain spreads:

Pixel Panel

EDIT:
Heres the patch too! Sorry its a bit messy!

PixelMap.v4p (51.4 kB)

since you’re already using pipet and have direct access to all the colors in vvvv, you could use RGB (Color Split) , then Map (Value) , Map (Value) Interval or similar and finally RGB (Color Join) to do the color correction yourself.

Hi Diki thanks for that,
just tried it now with an RGB node as you said and an HSL but still cant seem to get good color difference. Ill try and map this exponentially and see if that gives the desired result, els i may try and process the source material in after effects or something.

Cheers

try the gamma node…

Might be better to add gamma curve to it as human eye vision isnt that plain as Leds are. Human eye sees brightness more or less in logarithm steps. Leds only have that plain 0-255. As a result you see only a few contrast steps, depending of your overall brightness of Led. Contrast=BrightestPoint/DarkestPoint.
Moreover, we see more contrasts in lower values/darkness. Thats why you think midtones are not visible. You just see lower tones and glaring high values. (Look how your camera sees brightness different than you do)

Thats why expensive color changers do a brightness fade by wedge-shaped neutral grey glass, fading from opaque to clear. You are not bound to only 255 brightness values, which are less steps in human vision (or to be precise, you have 255 steps for the drive to move the wedge).

See here for that logarithm stuff

So, as result there isnt a satisfying solution for your problem in terms that you have only a few visible contrast steps.

edit: catweasel was faster :)

maybe something like this too

Pic_contrast.v4p (20.4 kB)

Whoah! Thanks guys, looks like a whole new avenue of learning here!
@Colorsound, thanks for the patch, it looks awesome!
I’ll see what results i get when i have some time tonight.