Resolution drops in motion? at DVinfo.net

Mike Thomann · March 31st, 2008, 02:34 AM

I've read in a few places now that that the EX1 loses resolution in motion. Here's one of the more descriptive explanations of someone's experience...

"the EX1 can deliver a noticeably sharper image especially on a static shot. In 720p mode they're very comparable, in 1080 mode the EX1 pulls significantly ahead in terms of sharpness -- as long as the camera isn't moving. In motion the resolution drops off a lot, and they end up being about the same on a pan. I don't understand the exact reasons why this happens (although I have my suspicions) but it's not something I'd blame on MPEG-2; it happens on the live output. In fact, it happens on the cam's own LCD -- go to 100% wide angle, turn on peaking (especially in yellow) and shoot some trees, etc. You should see a sea of yellow peaking. Now pan side to side, and all your peaking disappears until you stop moving. That doesn't happen on an HVX or, well, any other camera I can think of, but it happens on the Sony."

"My theory is that they're doing some aggressive noise reduction algorithms, averaging frames together to bring the noise down, and moving the camera averages out that fine detail. It results in great-looking static shots but the shots go softer in motion. (btw, this actually helps the MPEG encoding, so I'm not saying it's a bad design decision, but just be aware that the super-sharp frame grabs you're going to see from this camera don't necessarily translate into super-sharp motion footage)."

...Notice they say they haven't seen this in other cameras. Any thoughts or experiences anyone?

Alexander Kubalsky · March 31st, 2008, 03:06 AM

There was alot of hysteria about this subject a few months ago when a certain HVX guru gave a somewhat dramatic review of the EX1.

I dont know much bout nothin but in a shop last week I had a EX1 and a HVX (with an external monitor attached) right in front of me and I did some slow pans with both set at 24p, The level of motion blur looked identical to me. Then I switched the EX1 to 30p and panned again, and noticed significantly less motion blur , same deal on the HVX.

I then stood still, looking forward, panned my head across the room and found that my eyes too had motion blur. Then I realized I really should get out more.

lol.

Matt Davis · March 31st, 2008, 04:38 AM

Quote:

Originally Posted by Mike Thomann

Any thoughts or experiences anyone

I would expect the digital peaking indicator to disappear when doing a pan, partly to do with motion blur (Doh!) and partly because there's a slight processing lag where the computer decides where to put the coloured pixels.

Steven Thomas · March 31st, 2008, 06:38 AM

That's from Barry's post a couple months ago.
Yes, this topic went on for a bit.

Until, somebody actually shot rez charts on motion.... No problem.

Read the whole thread.

Sean Donnelly · March 31st, 2008, 07:06 AM

Try dropping the shutter angle down to 11 degrees. You'll get some pretty sharp motion that way. It's a function of exposure time related to camera movement. The motion blur does of course help the MPEG 2 compression, but it happens on film as well. This is yet another reason safe panning speed guidelines came about long ago.

-Sean

Eric Pascarelli · March 31st, 2008, 07:15 AM

That claim has been thoroughly debunked. It's completely false.

The tester was using a waveform monitor to measure resolution, which aggregates all of the video lines (spatially) into one display. Because of the slight skew from the rolling shutter when slowly panning, all of the lines combined into one are slighty offset horizontally, giving the appearance of softness on the scope.

It's just a case of the wrong tool being used to measure. Instead of looking at an aggregate of all of the lines, the tester should have been looking at a single line to judge resolution on the scope.

Slightly off-topic, but a note on EX1 peaking:

It's odd but true, but peaking on the EX1 seems to look for contrast only horizontally and not vertically. So even if motion blur causes a momentary loss in the peaking readout, it only occurs when panning, not tilting.

If you put a res chart in front of a static EX1, you'll see lots of peaking displayed around groups of vertical lines but none around groups of horizontal lines.

Sean Donnelly · March 31st, 2008, 07:25 AM

That's interesting Eric, I guess it's because of the scan direction of the shutter since it would have to stack all the lines and compare them vertically, or it could just read out line by line directly.

-Sean

Anmol Mishra · April 8th, 2008, 07:31 AM

Any link to this thread ?

Quote:

Originally Posted by Steven Thomas

That's from Barry's post a couple months ago.
Yes, this topic went on for a bit.

Until, somebody actually shot rez charts on motion.... No problem.

Read the whole thread.

Barry Green · April 9th, 2008, 11:37 AM

The "softening during motion" was indeed related to the way the rolling shutter interacted with the waveform monitor (and the way the camera's peaking circuit works). Turns out we thought we were seeing something but after testing it in a different way (using a slanted series of vertical bars to "correct" for the skew during motion) we were able to verify that it's not actually losing resolution; it just affected the waveform in a way to make it look as if it was.

March 31st, 2008, 02:34 AM	#1
Mike Thomann Regular Crew Join Date: May 2007 Location: Amherst, NY Posts: 100	Resolution drops in motion? I've read in a few places now that that the EX1 loses resolution in motion. Here's one of the more descriptive explanations of someone's experience... "the EX1 can deliver a noticeably sharper image especially on a static shot. In 720p mode they're very comparable, in 1080 mode the EX1 pulls significantly ahead in terms of sharpness -- as long as the camera isn't moving. In motion the resolution drops off a lot, and they end up being about the same on a pan. I don't understand the exact reasons why this happens (although I have my suspicions) but it's not something I'd blame on MPEG-2; it happens on the live output. In fact, it happens on the cam's own LCD -- go to 100% wide angle, turn on peaking (especially in yellow) and shoot some trees, etc. You should see a sea of yellow peaking. Now pan side to side, and all your peaking disappears until you stop moving. That doesn't happen on an HVX or, well, any other camera I can think of, but it happens on the Sony." "My theory is that they're doing some aggressive noise reduction algorithms, averaging frames together to bring the noise down, and moving the camera averages out that fine detail. It results in great-looking static shots but the shots go softer in motion. (btw, this actually helps the MPEG encoding, so I'm not saying it's a bad design decision, but just be aware that the super-sharp frame grabs you're going to see from this camera don't necessarily translate into super-sharp motion footage)." ...Notice they say they haven't seen this in other cameras. Any thoughts or experiences anyone?

March 31st, 2008, 07:15 AM	#6
Eric Pascarelli Major Player Join Date: Nov 2007 Location: Los Angeles, CA Posts: 498	No resolution loss when moving - completely false That claim has been thoroughly debunked. It's completely false. The tester was using a waveform monitor to measure resolution, which aggregates all of the video lines (spatially) into one display. Because of the slight skew from the rolling shutter when slowly panning, all of the lines combined into one are slighty offset horizontally, giving the appearance of softness on the scope. It's just a case of the wrong tool being used to measure. Instead of looking at an aggregate of all of the lines, the tester should have been looking at a single line to judge resolution on the scope. Slightly off-topic, but a note on EX1 peaking: It's odd but true, but peaking on the EX1 seems to look for contrast only horizontally and not vertically. So even if motion blur causes a momentary loss in the peaking readout, it only occurs when panning, not tilting. If you put a res chart in front of a static EX1, you'll see lots of peaking displayed around groups of vertical lines but none around groups of horizontal lines.

March 31st, 2008, 03:06 AM	#2
Alexander Kubalsky Regular Crew Join Date: Jan 2008 Location: Japan Posts: 172	There was alot of hysteria about this subject a few months ago when a certain HVX guru gave a somewhat dramatic review of the EX1. I dont know much bout nothin but in a shop last week I had a EX1 and a HVX (with an external monitor attached) right in front of me and I did some slow pans with both set at 24p, The level of motion blur looked identical to me. Then I switched the EX1 to 30p and panned again, and noticed significantly less motion blur , same deal on the HVX. I then stood still, looking forward, panned my head across the room and found that my eyes too had motion blur. Then I realized I really should get out more. lol.

March 31st, 2008, 06:38 AM	#4
Steven Thomas Trustee Join Date: Sep 2005 Location: Gilbert, AZ Posts: 1,896	That's from Barry's post a couple months ago. Yes, this topic went on for a bit. Until, somebody actually shot rez charts on motion.... No problem. Read the whole thread.

March 31st, 2008, 07:06 AM	#5
Sean Donnelly Regular Crew Join Date: Jan 2008 Location: New York, NY Posts: 188	Try dropping the shutter angle down to 11 degrees. You'll get some pretty sharp motion that way. It's a function of exposure time related to camera movement. The motion blur does of course help the MPEG 2 compression, but it happens on film as well. This is yet another reason safe panning speed guidelines came about long ago. -Sean

March 31st, 2008, 07:25 AM	#7
Sean Donnelly Regular Crew Join Date: Jan 2008 Location: New York, NY Posts: 188	That's interesting Eric, I guess it's because of the scan direction of the shutter since it would have to stack all the lines and compare them vertically, or it could just read out line by line directly. -Sean

April 9th, 2008, 11:37 AM	#9
Barry Green Barry Wan Kenobi Join Date: Jul 2003 Location: North Carolina Posts: 3,863	The "softening during motion" was indeed related to the way the rolling shutter interacted with the waveform monitor (and the way the camera's peaking circuit works). Turns out we thought we were seeing something but after testing it in a different way (using a slanted series of vertical bars to "correct" for the skew during motion) we were able to verify that it's not actually losing resolution; it just affected the waveform in a way to make it look as if it was.