Pink 'edge' on footage at DVinfo.net

Alex Taylor · December 12th, 2002, 09:05 PM

Forgive me if this is the wrong forum..

In the quest for cheap lighting solutions I decided to try those yellow construction lights with the two 500w halogens mounted on them. I set them up in my room and shot some footage and it looked great. I reviewed it on the TV and then brought it into Premiere to try playing around with colour correction, and it still looked good. I decided I'd bring some of the footage into After Effects, and all of a sudden there is a visible pink 'edge' on my shoulder (me being the subject of this footage). However, it's not AS visible in Premiere or on a TV, but a tiny bit of it is still there.

I'm guessing this has something to do with the camera's CCD? It's a Canon ZR20 1-chipper.

Here are some pics for reference:

That's what it looks like in AE..

And this is a frame grab from Premiere.

Rob Lohman · December 13th, 2002, 05:19 AM

Hmmm, it is strange that AE would render it differently. Very
strange. I do think this originates from the fact that your camera
only has 1 chip.

Jeff Donald · December 13th, 2002, 08:59 AM

I believe it to be an artifact of compression algorithms. In the video domain, a sharp edge represents a high frequency, or rate of change, and the faster the rate of change (sharp difference as between black and white), the more information there is in a picture. The job of a compression algorithm is to reduce transmitted or stored information, and sharp edges and sudden transients are the first to suffer. The different programs use different algorithms to do the compression and expansion, thus creating different artifacts.

A 3 chip camera can better handle the sharp transients and the artifacts are reduced. But artifacts are still present in most prosumer cameras.

Jeff

Barry Goyette · December 13th, 2002, 09:02 AM

The fact that it is occurring along the highest contrast edges would indicate some kind of chromatic aberation---which means its a combination of CCD and lens causing the problem. It is common for all CCD's to produce these kinds of defects...although this one looks pretty severe. I would test the camera in a variety of lighting situations (ambient and with your new lights)...to see when and how regularly it occurs. That way you can steer clear of it if it bothers you.

I doubt that the camera is out of spec...this effect is typical of small chip DV cams. My guess on why its worse in AE is that the screen grab is deinterlaced....perhaps this is adding to the "glow".

Barry

Andre De Clercq · December 13th, 2002, 09:22 AM

Was there any (even small) motion in the footage?. Can you show the frame before and after? Does your camcorder have an EIS which was on?

Alex Taylor · December 13th, 2002, 09:53 AM

Thanks for your replies!

Quote:

Was there any (even small) motion in the footage?. Can you show the frame before and after? Does your camcorder have an EIS which was on?

There was a very small bit of motion, the pink edge appears in all frames.

I just tried reversing the field order in AE and the pink edge turned to a more orange colour. I also duplicated the clip, reversed the field order on the top clip and made its opacity 50% much like the film look technique and it disappeared for the most part.

And the ZR20 has image stabilization, I'm not 100% sure whether its EIS or not. This afternoon I'll try some more tests and turn it off.

More images:

3 consecutive frame grabs from AE.
http://www.alextaylor.org/images/0;00;01;11.jpg
http://www.alextaylor.org/images/0;00;01;12.jpg
http://www.alextaylor.org/images/0;00;01;13.jpg

Two copies of the same footage with field order reversed and opacity at 50% on the top clip.
http://www.alextaylor.org/images/footage-doubled.jpg

Andre De Clercq · December 13th, 2002, 01:19 PM

Because single CCD chips have color filters (3 or 4 color strucures), the color sensors (pixels) are physically separated. This color spreading is is pretty well clustered by the signal processing if there is no motion, and becomes visually averaged with fast or normal motion. If there is a (very)small amount of motion specifically in vertical direction a kind of temporal aliasing can occur so that two (or more) consecutieve fields "stick" on the same color pixel combination and enhance the coresponding colors. This phenomenon belongs to the "dynamic color fringing" which is known for single CCD (and pixel shift) structures when slow motion or small shakes occur. Field order change and specific codec properties can change this behaviour. EIS too can influence (depending on the compensation delays) this problem.

December 12th, 2002, 09:05 PM	#1
Alex Taylor Major Player Join Date: Dec 2002 Location: Vancouver, BC Posts: 390	Pink 'edge' on footage Forgive me if this is the wrong forum.. In the quest for cheap lighting solutions I decided to try those yellow construction lights with the two 500w halogens mounted on them. I set them up in my room and shot some footage and it looked great. I reviewed it on the TV and then brought it into Premiere to try playing around with colour correction, and it still looked good. I decided I'd bring some of the footage into After Effects, and all of a sudden there is a visible pink 'edge' on my shoulder (me being the subject of this footage). However, it's not AS visible in Premiere or on a TV, but a tiny bit of it is still there. I'm guessing this has something to do with the camera's CCD? It's a Canon ZR20 1-chipper. Here are some pics for reference: That's what it looks like in AE.. And this is a frame grab from Premiere. __________________ alextaylor.org

December 13th, 2002, 05:19 AM	#2
Rob Lohman RED Code Chef Join Date: Oct 2001 Location: Holland Posts: 12,514	Hmmm, it is strange that AE would render it differently. Very strange. I do think this originates from the fact that your camera only has 1 chip. __________________ Rob Lohman, visuar@iname.com DV Info Wrangler & RED Code Chef Join the DV Challenge \| Lady X Search DVinfo.net for quick answers \| Buy from the best: DVinfo.net sponsors

December 13th, 2002, 08:59 AM	#3
Jeff Donald Warden Join Date: Mar 2002 Location: Clearwater, FL Posts: 8,287	I believe it to be an artifact of compression algorithms. In the video domain, a sharp edge represents a high frequency, or rate of change, and the faster the rate of change (sharp difference as between black and white), the more information there is in a picture. The job of a compression algorithm is to reduce transmitted or stored information, and sharp edges and sudden transients are the first to suffer. The different programs use different algorithms to do the compression and expansion, thus creating different artifacts. A 3 chip camera can better handle the sharp transients and the artifacts are reduced. But artifacts are still present in most prosumer cameras. Jeff

December 13th, 2002, 09:02 AM	#4
Barry Goyette Trustee Join Date: Oct 2001 Location: San Luis Obispo CA Posts: 1,195	The fact that it is occurring along the highest contrast edges would indicate some kind of chromatic aberation---which means its a combination of CCD and lens causing the problem. It is common for all CCD's to produce these kinds of defects...although this one looks pretty severe. I would test the camera in a variety of lighting situations (ambient and with your new lights)...to see when and how regularly it occurs. That way you can steer clear of it if it bothers you. I doubt that the camera is out of spec...this effect is typical of small chip DV cams. My guess on why its worse in AE is that the screen grab is deinterlaced....perhaps this is adding to the "glow". Barry

December 13th, 2002, 09:22 AM	#5
Andre De Clercq Major Player Join Date: Sep 2002 Location: Belgium Posts: 804	Was there any (even small) motion in the footage?. Can you show the frame before and after? Does your camcorder have an EIS which was on?

December 13th, 2002, 01:19 PM	#7
Andre De Clercq Major Player Join Date: Sep 2002 Location: Belgium Posts: 804	Because single CCD chips have color filters (3 or 4 color strucures), the color sensors (pixels) are physically separated. This color spreading is is pretty well clustered by the signal processing if there is no motion, and becomes visually averaged with fast or normal motion. If there is a (very)small amount of motion specifically in vertical direction a kind of temporal aliasing can occur so that two (or more) consecutieve fields "stick" on the same color pixel combination and enhance the coresponding colors. This phenomenon belongs to the "dynamic color fringing" which is known for single CCD (and pixel shift) structures when slow motion or small shakes occur. Field order change and specific codec properties can change this behaviour. EIS too can influence (depending on the compensation delays) this problem.