Kerr Cooks first look at FX1 at DVinfo.net

Michael Pappas · October 24th, 2004, 12:56 AM

I found this excerpt to be very disturbing from Kerr Cook. Link to whole first FX1 look below.

<<<With a slight movement (of the whole scene), the resolution immediately drops and details are lost. The "sharpness" and details appear immediately when the panning/movement stops. If I just shake the camcorder back and forth, I could see the "blurriness" or loss of sharp detail. I guess this was done to reduce the information that MPEG2 needs to encode (when panning or when the entire scene is changing). I recorded some trees (without leaves) against the sky. When panning, the thin branches disappear until I stop and they "pop" into view.>>>>

LINK: http://www.sonyhdvinfo.com/showthread.php?t=348

Michael Pappas · October 24th, 2004, 01:05 AM

Kerr cook, took the camera to a bestbuy and viewed the material on a HD monitor and said that the issue was not as bad as it was in the VF. Could be good news....

Lynne Whelden · October 24th, 2004, 06:40 AM

The good Dr. Steve Mullen, who used to moderate this site, arrived at the same conclusion. I quote from the latest issue of "VideoSystems," "...with pixel offset technology, effective horizontal resolution is a function of the COLORS, the COLOR PATTERNS, and the MOTION of colored objects in a scene." In other words, the resolution is never fixed but depends on what's in a scene!!!
Thus I see at least 3 ways Sony has made this camera deliberately inferior:
1--using pixel offset--variable resolution
2--using MPEG-1 Layer 2 for audio--lossy
3--its sheer weight of over 5 pounds--heavy and awkward

Donal Briard · October 24th, 2004, 08:20 AM

'2--using MPEG-1 Layer 2 for audio--lossy"

That is the HDV standard for audio.

Mark Grant · October 24th, 2004, 08:36 AM

"using pixel offset--variable resolution"

No-one seems to complain about the XL1 (and presumably the XL2?) using pixel shift. I'd prefer a full 1440 pixel CCD, but if that would add $5k to the price, I'll live with pixel shift.

"That is the HDV standard for audio."

And provided you don't keep recompressing it in the edit and don't record footage that really pushes the limitations of the compression, no-one is going to notice the difference. For recording dialog and the like, it will be fine in 99.9% of cases.

My main concern is still the reported issues with pans: I haven't noticed anything really annoying with the few fast pans in the footage posted here, though they do look a little odd at times.

Michael Pappas · October 24th, 2004, 11:07 AM

There is a difference between "pixel shift" and "variable resolution". Pixel shift is used on broadcast cameras etc. Nothing wrong with it at all. Variable resolution is an entirely different technology. This is where your cameras resolution adjusts depending on the content detail level that you are filming. Pixel shift does one thing and never deviates from the that. A resolution that floats around to a point where you can see it may be an issue. I am waiting for good content from the FX1 to judge more.

Michael Pappas

Mark Grant · October 24th, 2004, 11:12 AM

And where does that article say anything about the camera varying resolution? It's talking about pixel shift (and its limitations), not some kind of DSP-based resolution change.

Barry Green · October 24th, 2004, 03:35 PM

The lower-detail-when-panning is present in the JVC HD1 and HD10. I have no doubt it's in the FX1 as well.

I'd wager it's not anything Sony's doing on purpose, and most certainly not something done to deliberately cripple the camera. It's going to be an artifact of the MPEG-2 compression system.

MPEG-2 compresses most efficiently when there's little change between frames. The more change there is between frames, the less efficent MPEG-2 is. So on pans, where every pixel is changing, you're going to see less detail, because you're going to see the effect of higher compression ratios. On static shots you should see incredible resolution, because the less that changes between scenes, the less that has to be re-compressed.

Like it or not, HDV is MPEG-2. MPEG-2 is used in HDV. There will never be an HDV camera that doesn't use MPEG-2; speculation to the contrary is wasted breath. If the camera says HDV on it, it uses MPEG-2, and this is an artifact of MPEG-2. An unavoidable artifact.

Maybe one reason why Panasonic has held off on embracing HDV, and maybe their eventual entry will be based on DVCPRO-HD, which (as I understand it) is a derivative of frame-based compression, where every frame is compressed individually (like DV and MJPG).

Michael Pappas · October 24th, 2004, 03:45 PM

<<<-- Originally posted by Mark Grant : And where does that article say anything about the camera varying resolution? It's talking about pixel shift (and its limitations), not some kind of DSP-based resolution change. -->>>

Mark Grant , Kerr says it right here. Just as I posted a few post above. What's the problem mark? And in the quote above and reposted below there is no mention of "pixel shift".

Repost from above for Mark to re-read..........
<<<With a slight movement (of the whole scene), the resolution immediately drops and details are lost. The "sharpness" and details appear immediately when the panning/movement stops. If I just shake the camcorder back and forth, I could see the "blurriness" or loss of sharp detail. I guess this was done to reduce the information that MPEG2 needs to encode (when panning or when the entire scene is changing). I recorded some trees (without leaves) against the sky. When panning, the thin branches disappear until I stop and they "pop" into view.>>>>

Michael Pappas · October 24th, 2004, 03:54 PM

Mark, I see your responding to Mullens quote I think. I started this thread with a quote of Kerrs review. Someone stuck something from mullen in the thread. I am looking at Kerrs quote and saying" there is no mention of pixel shift" Now I see what you were reading, oopps!

Cieonatechori Michelieve · October 24th, 2004, 03:55 PM

<<<-- When panning, the thin branches disappear until I stop and they "pop" into view.>>>> -->>>

but later he says that when he saw it on HDTV, the branches were visible. it's only on camera lcd where he cudnt see them

Mark Grant · October 24th, 2004, 05:18 PM

"Mark, I see your responding to Mullens quote I think"

Yes, and he was pointing out that _all_ pixel shift cameras will suffer from lower resolution in unusual circumstances: e.g. if you have a scene that's mostly green with little red and blue, or a scene that's mostly red and blue with little green. But that's just a limitation of the pixel shift design, it's not a deliberate attempt to reduce resolution.

Equally, while the one fast pan I've seen in the footage posted here looks a little odd, I very much doubt that Sony are deliberately reducing resolution on pans: far more likely it's just the limitation of the MPEG-2 compression.

Michael Pappas · October 24th, 2004, 05:22 PM

kerr Cook posted about two hours later after the above quote that he took it to bestbuy and saw that it didn't.

Kerr Cook wrote:
<<,The lower resolution was noticable when moving, but NOT AS MUCH on the real TV as on the LCD panel when I was recording. E.g. the thin tree branches disappeared from view on the LCD when panning or dithering but showed on the TV.>>>

This could be more of a motion blur then mpeg2 on playback. Since HD is more precise, blurring is going to be more visible just like bad focus. Just a theory!

Michael Pappas · October 24th, 2004, 05:26 PM

Why do my Kaku clips look like 30P video as apposed to my live 60 field look. I have HD NTSC 60i clips that play back with the live video look. Kaku's material looks like frame mode/30P like my Iwerks films and DVX100 30p mode material.

October 24th, 2004, 12:56 AM	#1
Michael Pappas Regular Crew Join Date: Sep 2001 Location: California Posts: 667	Kerr Cooks first lok at FX1 I found this excerpt to be very disturbing from Kerr Cook. Link to whole first FX1 look below. <<<With a slight movement (of the whole scene), the resolution immediately drops and details are lost. The "sharpness" and details appear immediately when the panning/movement stops. If I just shake the camcorder back and forth, I could see the "blurriness" or loss of sharp detail. I guess this was done to reduce the information that MPEG2 needs to encode (when panning or when the entire scene is changing). I recorded some trees (without leaves) against the sky. When panning, the thin branches disappear until I stop and they "pop" into view.>>>> LINK: http://www.sonyhdvinfo.com/showthread.php?t=348

October 24th, 2004, 06:40 AM	#3
Lynne Whelden Major Player Join Date: Feb 2003 Location: williamsport, pa Posts: 604	How Sony crippled this camera... The good Dr. Steve Mullen, who used to moderate this site, arrived at the same conclusion. I quote from the latest issue of "VideoSystems," "...with pixel offset technology, effective horizontal resolution is a function of the COLORS, the COLOR PATTERNS, and the MOTION of colored objects in a scene." In other words, the resolution is never fixed but depends on what's in a scene!!! Thus I see at least 3 ways Sony has made this camera deliberately inferior: 1--using pixel offset--variable resolution 2--using MPEG-1 Layer 2 for audio--lossy 3--its sheer weight of over 5 pounds--heavy and awkward

October 24th, 2004, 01:05 AM	#2
Michael Pappas Regular Crew Join Date: Sep 2001 Location: California Posts: 667	Kerr cook, took the camera to a bestbuy and viewed the material on a HD monitor and said that the issue was not as bad as it was in the VF. Could be good news....

October 24th, 2004, 08:20 AM	#4
Donal Briard Regular Crew Join Date: Sep 2004 Posts: 43	'2--using MPEG-1 Layer 2 for audio--lossy" That is the HDV standard for audio.

October 24th, 2004, 08:36 AM	#5
Mark Grant Regular Crew Join Date: Oct 2004 Location: Saskatoon, Canada (was London, UK) Posts: 138	"using pixel offset--variable resolution" No-one seems to complain about the XL1 (and presumably the XL2?) using pixel shift. I'd prefer a full 1440 pixel CCD, but if that would add $5k to the price, I'll live with pixel shift. "That is the HDV standard for audio." And provided you don't keep recompressing it in the edit and don't record footage that really pushes the limitations of the compression, no-one is going to notice the difference. For recording dialog and the like, it will be fine in 99.9% of cases. My main concern is still the reported issues with pans: I haven't noticed anything really annoying with the few fast pans in the footage posted here, though they do look a little odd at times.

October 24th, 2004, 11:07 AM	#6
Michael Pappas Regular Crew Join Date: Sep 2001 Location: California Posts: 667	There is a difference between "pixel shift" and "variable resolution". Pixel shift is used on broadcast cameras etc. Nothing wrong with it at all. Variable resolution is an entirely different technology. This is where your cameras resolution adjusts depending on the content detail level that you are filming. Pixel shift does one thing and never deviates from the that. A resolution that floats around to a point where you can see it may be an issue. I am waiting for good content from the FX1 to judge more. Michael Pappas

October 24th, 2004, 11:12 AM	#7
Mark Grant Regular Crew Join Date: Oct 2004 Location: Saskatoon, Canada (was London, UK) Posts: 138	And where does that article say anything about the camera varying resolution? It's talking about pixel shift (and its limitations), not some kind of DSP-based resolution change.

October 24th, 2004, 03:35 PM	#8
Barry Green Barry Wan Kenobi Join Date: Jul 2003 Location: North Carolina Posts: 3,863	The lower-detail-when-panning is present in the JVC HD1 and HD10. I have no doubt it's in the FX1 as well. I'd wager it's not anything Sony's doing on purpose, and most certainly not something done to deliberately cripple the camera. It's going to be an artifact of the MPEG-2 compression system. MPEG-2 compresses most efficiently when there's little change between frames. The more change there is between frames, the less efficent MPEG-2 is. So on pans, where every pixel is changing, you're going to see less detail, because you're going to see the effect of higher compression ratios. On static shots you should see incredible resolution, because the less that changes between scenes, the less that has to be re-compressed. Like it or not, HDV is MPEG-2. MPEG-2 is used in HDV. There will never be an HDV camera that doesn't use MPEG-2; speculation to the contrary is wasted breath. If the camera says HDV on it, it uses MPEG-2, and this is an artifact of MPEG-2. An unavoidable artifact. Maybe one reason why Panasonic has held off on embracing HDV, and maybe their eventual entry will be based on DVCPRO-HD, which (as I understand it) is a derivative of frame-based compression, where every frame is compressed individually (like DV and MJPG).

October 24th, 2004, 03:45 PM	#9
Michael Pappas Regular Crew Join Date: Sep 2001 Location: California Posts: 667	<<<-- Originally posted by Mark Grant : And where does that article say anything about the camera varying resolution? It's talking about pixel shift (and its limitations), not some kind of DSP-based resolution change. -->>> Mark Grant , Kerr says it right here. Just as I posted a few post above. What's the problem mark? And in the quote above and reposted below there is no mention of "pixel shift". Repost from above for Mark to re-read.......... <<<With a slight movement (of the whole scene), the resolution immediately drops and details are lost. The "sharpness" and details appear immediately when the panning/movement stops. If I just shake the camcorder back and forth, I could see the "blurriness" or loss of sharp detail. I guess this was done to reduce the information that MPEG2 needs to encode (when panning or when the entire scene is changing). I recorded some trees (without leaves) against the sky. When panning, the thin branches disappear until I stop and they "pop" into view.>>>>

October 24th, 2004, 03:54 PM	#10
Michael Pappas Regular Crew Join Date: Sep 2001 Location: California Posts: 667	Mark, I see your responding to Mullens quote I think. I started this thread with a quote of Kerrs review. Someone stuck something from mullen in the thread. I am looking at Kerrs quote and saying" there is no mention of pixel shift" Now I see what you were reading, oopps!

October 24th, 2004, 03:55 PM	#11
Cieonatechori Michelieve Regular Crew Join Date: Aug 2004 Location: Washington DC Posts: 25	<<<-- When panning, the thin branches disappear until I stop and they "pop" into view.>>>> -->>> but later he says that when he saw it on HDTV, the branches were visible. it's only on camera lcd where he cudnt see them

October 24th, 2004, 05:18 PM	#12
Mark Grant Regular Crew Join Date: Oct 2004 Location: Saskatoon, Canada (was London, UK) Posts: 138	"Mark, I see your responding to Mullens quote I think" Yes, and he was pointing out that _all_ pixel shift cameras will suffer from lower resolution in unusual circumstances: e.g. if you have a scene that's mostly green with little red and blue, or a scene that's mostly red and blue with little green. But that's just a limitation of the pixel shift design, it's not a deliberate attempt to reduce resolution. Equally, while the one fast pan I've seen in the footage posted here looks a little odd, I very much doubt that Sony are deliberately reducing resolution on pans: far more likely it's just the limitation of the MPEG-2 compression.

October 24th, 2004, 05:22 PM	#13
Michael Pappas Regular Crew Join Date: Sep 2001 Location: California Posts: 667	kerr Cook posted about two hours later after the above quote that he took it to bestbuy and saw that it didn't. Kerr Cook wrote: <<,The lower resolution was noticable when moving, but NOT AS MUCH on the real TV as on the LCD panel when I was recording. E.g. the thin tree branches disappeared from view on the LCD when panning or dithering but showed on the TV.>>> This could be more of a motion blur then mpeg2 on playback. Since HD is more precise, blurring is going to be more visible just like bad focus. Just a theory!

October 24th, 2004, 05:26 PM	#14
Michael Pappas Regular Crew Join Date: Sep 2001 Location: California Posts: 667	Why do my Kaku clips look like 30P video as apposed to my live 60 field look. I have HD NTSC 60i clips that play back with the live video look. Kaku's material looks like frame mode/30P like my Iwerks films and DVX100 30p mode material.