25% drop in res? at DVinfo.net

Thomas Fraser · March 12th, 2005, 10:23 PM

I was told that the GL2 resolution drops by 25% when you go into frame mode. Is this true?

Boyd Ostroff · March 12th, 2005, 10:35 PM

This has frequently been discussed in the past, mostly in relation to the XL-1, but I think you'll find the issues are the same:

http://www.dvinfo.net/conf/showthrea...threadid=14726
http://www.dvinfo.net/conf/showthrea...threadid=35158
http://www.dvinfo.net/conf/showthrea...=&threadid=983
http://www.dvinfo.net/conf/showthrea...&threadid=2956

Use our seach feature to help find answers to questions like this.

Steve Brady · March 20th, 2005, 06:13 AM

There doesn't seem to be a lot of solid information in any of that lot, though.

I've set my XM2 up on a tripod in front of a resolution chart and used the remote to switch between normal and frame modes, and it does indeed seem as though there's a resolution loss of 25% - I eyeball about 440 lines of vertical res in normal mode versus about 330 lines in frame mode.

What's missing is the explanation why. There's some reference to "interpolation" and "blending", but I don't think it's based on any solid technical information. It's an academic point, of course, because there probably isn't anything that can be done about it, but some of us just like to know these things.

Rob Lohman · March 20th, 2005, 06:22 AM

The reason this happens is that the GL1/2 and XL1(S) line of
camera's does not have a true progressive mode. This frame
mode uses a special algorithm and CCD accessing to achieve
the progressive look. It is better (in my opinion) than a lot of
de-interlacing programs/algorithms, but you indeed loose some
resolution.

Personally I don't care about this. It's the story and entertaining
value of my movie that should hold viewers. Not the amount of
lines of resolution it has.

As I always say: if it looks good enough it to your eyes....

David Ennis · March 20th, 2005, 11:10 AM

Seems odd, then, that the GL2 manual (p. 50) says,
"The frame mode can be used when you wish to look at a video scene by scene, free from field slippage........The vertical resolution of the image is improved by 1.5x. During playback pause the video at any time if you would like to see a crystal clear still image in the display or on a connected TV screen....."

Steve Brady · March 20th, 2005, 05:13 PM

No, that's right. A "still" in normal movie mode is a single field, which has only 50% of the resolution of a full frame. A still in frame movie mode is the full frame, which has 75% of the resolution of a normal movie mode frame. 75% is 150% of 50%.

It's still not clear to me why, though.

CCDs are always progressive. Usually, lines of video are constructed by averaging two successive rows of data from the CCD - for the upper field, the first and second rows are averaged to create the first line, the third and fourth rows are averaged to create the second line and so on. For the lower field, the second and third rows are averaged to create the first line, the fourth and fifth rows are averaged to create the second line, etc. This reduces twitter and improves the signal to noise ratio by 3dB. Some broadcast cameras allow you to control the ratio of successive rows to trade off noise and flicker against resolution.

Because the green CCD in the Canon cameras is offset from the red and blue CCDs by half a pixel ("pixel-shift"), I wonder whether the video lines in normal movie mode might be created by using a 25:75/75:25 split instead of 50:50. This will give a sharper vertical resolution, and re-align the green rows with the red and blue rows, at the expense of increased twitter and noise.

For frame movie mode, however, there will already be twice as much noise (because you can only use one read of the CCDs instead of two, so you effectively have half as much signal and the same amount of noise) and if the purpose of frame movie mode is to achieve "crystal clear" stills, then you'd obviously also want to minimise twitter.

It seems to me that, since there's no technical limitation that prevents frame movie mode from having the same resolution as normal movie mode, the reduced resolution might be a deliberate choice on the part of Canon's designers, in order to reduce noise and twitter.

March 12th, 2005, 10:23 PM	#1
Thomas Fraser Regular Crew Join Date: Jan 2003 Location: Toronto Ontario Canada Posts: 70	25% drop in res? I was told that the GL2 resolution drops by 25% when you go into frame mode. Is this true? __________________ Thanks for your help everyone.

March 20th, 2005, 06:22 AM	#4
Rob Lohman RED Code Chef Join Date: Oct 2001 Location: Holland Posts: 12,514	The reason this happens is that the GL1/2 and XL1(S) line of camera's does not have a true progressive mode. This frame mode uses a special algorithm and CCD accessing to achieve the progressive look. It is better (in my opinion) than a lot of de-interlacing programs/algorithms, but you indeed loose some resolution. Personally I don't care about this. It's the story and entertaining value of my movie that should hold viewers. Not the amount of lines of resolution it has. As I always say: if it looks good enough it to your eyes.... __________________ 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

March 20th, 2005, 11:10 AM	#5
David Ennis Fred Retread Join Date: Jul 2004 Location: Hartford, CT Posts: 1,227	Seems odd, then, that the GL2 manual (p. 50) says, "The frame mode can be used when you wish to look at a video scene by scene, free from field slippage........The vertical resolution of the image is improved by 1.5x. During playback pause the video at any time if you would like to see a crystal clear still image in the display or on a connected TV screen....." __________________ "Nothing in the world can take the place of persistence..." - Calvin Coolidge "My brain is wired to want to know how other things are wired." - Me

March 12th, 2005, 10:35 PM	#2
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	This has frequently been discussed in the past, mostly in relation to the XL-1, but I think you'll find the issues are the same: http://www.dvinfo.net/conf/showthrea...threadid=14726 http://www.dvinfo.net/conf/showthrea...threadid=35158 http://www.dvinfo.net/conf/showthrea...=&threadid=983 http://www.dvinfo.net/conf/showthrea...&threadid=2956 Use our seach feature to help find answers to questions like this.

March 20th, 2005, 06:13 AM	#3
Steve Brady Regular Crew Join Date: Feb 2005 Location: Rainham, Kent, UK Posts: 69	There doesn't seem to be a lot of solid information in any of that lot, though. I've set my XM2 up on a tripod in front of a resolution chart and used the remote to switch between normal and frame modes, and it does indeed seem as though there's a resolution loss of 25% - I eyeball about 440 lines of vertical res in normal mode versus about 330 lines in frame mode. What's missing is the explanation why. There's some reference to "interpolation" and "blending", but I don't think it's based on any solid technical information. It's an academic point, of course, because there probably isn't anything that can be done about it, but some of us just like to know these things.

March 20th, 2005, 05:13 PM	#6
Steve Brady Regular Crew Join Date: Feb 2005 Location: Rainham, Kent, UK Posts: 69	No, that's right. A "still" in normal movie mode is a single field, which has only 50% of the resolution of a full frame. A still in frame movie mode is the full frame, which has 75% of the resolution of a normal movie mode frame. 75% is 150% of 50%. It's still not clear to me why, though. CCDs are always progressive. Usually, lines of video are constructed by averaging two successive rows of data from the CCD - for the upper field, the first and second rows are averaged to create the first line, the third and fourth rows are averaged to create the second line and so on. For the lower field, the second and third rows are averaged to create the first line, the fourth and fifth rows are averaged to create the second line, etc. This reduces twitter and improves the signal to noise ratio by 3dB. Some broadcast cameras allow you to control the ratio of successive rows to trade off noise and flicker against resolution. Because the green CCD in the Canon cameras is offset from the red and blue CCDs by half a pixel ("pixel-shift"), I wonder whether the video lines in normal movie mode might be created by using a 25:75/75:25 split instead of 50:50. This will give a sharper vertical resolution, and re-align the green rows with the red and blue rows, at the expense of increased twitter and noise. For frame movie mode, however, there will already be twice as much noise (because you can only use one read of the CCDs instead of two, so you effectively have half as much signal and the same amount of noise) and if the purpose of frame movie mode is to achieve "crystal clear" stills, then you'd obviously also want to minimise twitter. It seems to me that, since there's no technical limitation that prevents frame movie mode from having the same resolution as normal movie mode, the reduced resolution might be a deliberate choice on the part of Canon's designers, in order to reduce noise and twitter.