Effect of more pixels ? at DVinfo.net

October 19th, 2002, 10:27 AM

I'm new to his site, and have a question on resolution, which hopefully someone can help me with :

For mini DV camcorders is there a limitation in the increase of resolution compared with the number of pixels in a CCD ?

For example for a 3 CCD camcorder with 1M pixel CCD's, is the resoution achieved, greater than one with 470K pixel CCD's ?

Or does the limitation of the DV standard mean that the extra information is not used ? I only refer to DV recording , not stills.

Thanks

Andy H

Chris Hurd · October 19th, 2002, 12:32 PM

Generally speaking, the fewer the number of pixels, the larger they are physically and thus gather more light, but with a softer image.

The higher number of pixels, resolution is increased but low-light performance falls way off. More pixels crammed onto a CCD mean they're physically smaller and thus gather less light.

In the long run, DV is DV at 720x480 pixels (or 720x576 pixels in PAL land). Hope this helps,

Andre De Clercq · October 19th, 2002, 12:52 PM

Ideal (theoretically) is an infinite number of pixels and just combine (convolution) them to get the 720x480 (ntsc) or 720x576 (pal) RGB rasters out of them. Why?....One will never get optical aliasing. Why is this not realistic? Besides the technological limits, the fill factor limits in IT CCD's would result in an unacceptabe low sensitivity. Furhermore there is no need to get that infinite number if the optical resolution is limited by diffraction, lens MTF, or optical anti aliasing filters. So, it's a matter of trade offs and for pro camcorders the use of a high performance optical antialiasing filters.

Don Palomaki · October 20th, 2002, 06:54 AM

Yes, ther is a hard limit on the resolution of a DV image, the 720 pixel line. For NTSC that translates to about 346k effective pixels in the image. But that sets the high limit. Other factors can reduce the resolution of a system including the quality (frequency response) of the analog electronics, the lens, and of course the CCD pixel count. With 3-CCD camcorders, DSP techniques like pixel shift can be use to form a good image with less than 346k pixels.

Note that 1-CCD camcorders use color filters over individual pixels to form the color image. So that a 3-CCD with 250k effective pixles per CCD is in many respects similar to a 1-CCD camcorder with, say, 500-750k pixels (although color saturation will be better in the 3-CCD camcorder)

If you read the fine print you may find that many megapixel camcorders use the megapixel in still mode only, and drop to on the order of 470k effective pixels for video.

Also, camcorders that use electronic image stabilization need many more pixels to implement EIS. Numbers like 770k pixels for EIS, or 450k pixels for OIS.

In summary, an effective video pixel count over, say, 450k in a 3-CCD OIS-equipped camcorder will add little if anything to the DV image resolution

Andre De Clercq · October 20th, 2002, 07:26 AM

Don, sorry but there is no hard limit in CCD pixel count. The sampling rate in DV is indeed 720 samples per line, but this can be combined out of less or more CCD pixels per line by simple convolution techniques. The more CCD pixels, the less aliasing effects for a given resolution. Like I wrote, the ideal is in that respect an infinite number of CCD elements, or your 720 elements combined with an ideal optical anti aliasing filter which doesn't exist. So one can never get 720 points per line (540TVL) at 100% amplitude without strong aliasing effects. See also http://www.microsoft.com/hwdev/archive/TVBROADCAST/ovrsamp.asp

October 21st, 2002, 12:58 PM

Thanks very much for the information.

Andy

Don Palomaki · October 22nd, 2002, 06:59 AM

I agree that the more CCD pixels, the better the potential image resolution, even for DV with the 720 pixels line. However, as pixel count goes up, diminishing returns set in.

In the end, with almost any DV camcorder on the market today, the significantly greater issue is talent in directing, shooting, and editing. Perhaps the major issue with camcorders is getting enough light on the scene so that the camcorder used can produce its best image.

Tylar Mccoy · November 1st, 2002, 11:57 AM

Thanks for this info. I am in the middle of going through a shitload of nonsense with onecall about my gl-1, but when (or if) i finaly get a non defective unit from them I think i will be happy with the cam. After paying for it i got a little worried when someone told me that the 270k pikel count was way low and I could get the same results with a 1ccd cam. This guy is telling me that i will only get about an 8% improvement With the gl-1 than a 1 chipper. Is there any truth behind this?

Jeff Donald · November 1st, 2002, 04:09 PM

Everybody wants to narrow down performance to one number. Do you shop for cars by only looking at the Horse Power? Then don't shop for cameras by only looking at the number of pixels. Much more than total pixels goes into determining picture quality. Dre... and Don mention many of them above, lenses, diffraction, MTF, etc. Picture quality can not be determined by specs on a piece of paper. Try going to a retailer and looking at the images different cameras produce. Compare single chip camers to three chip cameras. I think you'll see more than a 8% difference.

Jeff

Tylar Mccoy · November 1st, 2002, 04:52 PM

Thanks, I had an optura100mc for about 2 weeks before I sent it back to onecall. (the 100mc also had a hot pixel) I decided to credit the 100mc and get a gl-1. When the gl-1 got here It too had a hot pixel and I only used it for about 20 minutes before packing it back up and sending it back. I have been looking at the gl-1 for about a year, and now that I can afford it im getting screwed around by a dealer. I cant wait untill I get a non defective unit so I can stop spending my time on this nightmare and just enjoy the product that I paid for over a month ago.

I live in a rural area, So I would have to travel about 350 miles to find any stores high class enough to put 3 chippers on display, But for the short time I used the gl-1 I could notice a huge leap in quality. The most difference I noticed was shen I shot a blanket that is full of bright vivid colors. The same blanket shot with the 2 cams looks different for sure. I am a video newbie. Another thing I was reading is that the gl-1 is much worse in low light because of its 3ccds weakining the light. But I found the gl-1 worked better than the 100mc in low light, (in my living room filming my friends puppy)

Andre De Clercq · November 2nd, 2002, 04:21 AM

Where do you read an hear all those strange things, like the 8% "improvement", or the 3 CCD weakening the light performance? None of these statements is true.

Tylar Mccoy · November 2nd, 2002, 02:15 PM

The 8% came from a guy who was reading one of my posts here, and emailed me.
Heres part of his msg:

"As for me, any 3CCD camcorders under $2400 are crap. Side by side comparison....single chipper and 3 chippers are 92/100%. Will you put out $800 extra to pay for 8% different? Picture quality is about 8% different as compared to 1chip JVC GR DV3000U $1000 range."

The thing about the 3ccd's being worse in low light was something I read on a website, I cannot remember where, But it said something like, The Beam splits the light up, and as it splits the light onto the 3 different ccd's it get weakend, It said if you want good low light get a 1 chipper Because they are the most sensitive. Thanks for putting these myth's to rest for me.

Jeff Donald · November 2nd, 2002, 02:57 PM

Single chip cameras generally perform better in low light than 3 chip cameras for the reason you mentioned. The beam splitter, splits the light for each chip, Red, Green, Blue. The 8% sounds like just one persons opinion to me.

Jeff

Andre De Clercq · November 2nd, 2002, 03:09 PM

FYI camera's are not compaired in term of %. If all the parameters (sensitivity, resolution, lens qual;itiy...) would be the same (not realistic), single chippers allways show more aliasing effects (edge business, moire) and color fringeing. On the color splitting story, the guy posting this nonsence aparently doesn't know that a color filter in single CCD does the same in terms of allowing some colors only to some pixels. Worse even, a 3 CCD projects ALL the incoming light of a primary color on the related CCD, a single CCD filter allows 1/3 th of the incoming light (in case of a 3 color CCD filter) to specific pixels and absorbs the remaining light projected on those specific pixels. resulting in one of the lower sensitivity problems.

October 19th, 2002, 10:27 AM	#1
Andy uk1 Posts: n/a	Effect of more pixels ? I'm new to his site, and have a question on resolution, which hopefully someone can help me with : For mini DV camcorders is there a limitation in the increase of resolution compared with the number of pixels in a CCD ? For example for a 3 CCD camcorder with 1M pixel CCD's, is the resoution achieved, greater than one with 470K pixel CCD's ? Or does the limitation of the DV standard mean that the extra information is not used ? I only refer to DV recording , not stills. Thanks Andy H

October 19th, 2002, 12:32 PM	#2
Chris Hurd Obstreperous Rex Join Date: Jan 2001 Location: San Marcos, TX Posts: 27,366 Images: 513	Generally speaking, the fewer the number of pixels, the larger they are physically and thus gather more light, but with a softer image. The higher number of pixels, resolution is increased but low-light performance falls way off. More pixels crammed onto a CCD mean they're physically smaller and thus gather less light. In the long run, DV is DV at 720x480 pixels (or 720x576 pixels in PAL land). Hope this helps, __________________ CH Search DV Info Net \| 20 years of DVi \| ...Tuesday is Soylent Green Day!

October 19th, 2002, 12:52 PM	#3
Andre De Clercq Major Player Join Date: Sep 2002 Location: Belgium Posts: 804	Ideal (theoretically) is an infinite number of pixels and just combine (convolution) them to get the 720x480 (ntsc) or 720x576 (pal) RGB rasters out of them. Why?....One will never get optical aliasing. Why is this not realistic? Besides the technological limits, the fill factor limits in IT CCD's would result in an unacceptabe low sensitivity. Furhermore there is no need to get that infinite number if the optical resolution is limited by diffraction, lens MTF, or optical anti aliasing filters. So, it's a matter of trade offs and for pro camcorders the use of a high performance optical antialiasing filters.

October 20th, 2002, 06:54 AM	#4
Don Palomaki Wrangler Join Date: Sep 2001 Location: Northern VA Posts: 4,487	Yes, ther is a hard limit on the resolution of a DV image, the 720 pixel line. For NTSC that translates to about 346k effective pixels in the image. But that sets the high limit. Other factors can reduce the resolution of a system including the quality (frequency response) of the analog electronics, the lens, and of course the CCD pixel count. With 3-CCD camcorders, DSP techniques like pixel shift can be use to form a good image with less than 346k pixels. Note that 1-CCD camcorders use color filters over individual pixels to form the color image. So that a 3-CCD with 250k effective pixles per CCD is in many respects similar to a 1-CCD camcorder with, say, 500-750k pixels (although color saturation will be better in the 3-CCD camcorder) If you read the fine print you may find that many megapixel camcorders use the megapixel in still mode only, and drop to on the order of 470k effective pixels for video. Also, camcorders that use electronic image stabilization need many more pixels to implement EIS. Numbers like 770k pixels for EIS, or 450k pixels for OIS. In summary, an effective video pixel count over, say, 450k in a 3-CCD OIS-equipped camcorder will add little if anything to the DV image resolution

October 20th, 2002, 07:26 AM	#5
Andre De Clercq Major Player Join Date: Sep 2002 Location: Belgium Posts: 804	Don, sorry but there is no hard limit in CCD pixel count. The sampling rate in DV is indeed 720 samples per line, but this can be combined out of less or more CCD pixels per line by simple convolution techniques. The more CCD pixels, the less aliasing effects for a given resolution. Like I wrote, the ideal is in that respect an infinite number of CCD elements, or your 720 elements combined with an ideal optical anti aliasing filter which doesn't exist. So one can never get 720 points per line (540TVL) at 100% amplitude without strong aliasing effects. See also http://www.microsoft.com/hwdev/archive/TVBROADCAST/ovrsamp.asp

October 21st, 2002, 12:58 PM	#6
Andy uk1 Posts: n/a	Thanks very much for the information. Andy

October 22nd, 2002, 06:59 AM	#7
Don Palomaki Wrangler Join Date: Sep 2001 Location: Northern VA Posts: 4,487	I agree that the more CCD pixels, the better the potential image resolution, even for DV with the 720 pixels line. However, as pixel count goes up, diminishing returns set in. In the end, with almost any DV camcorder on the market today, the significantly greater issue is talent in directing, shooting, and editing. Perhaps the major issue with camcorders is getting enough light on the scene so that the camcorder used can produce its best image.

November 1st, 2002, 11:57 AM	#8
Tylar Mccoy New Boot Join Date: Sep 2002 Posts: 18	Thanks for this info. I am in the middle of going through a shitload of nonsense with onecall about my gl-1, but when (or if) i finaly get a non defective unit from them I think i will be happy with the cam. After paying for it i got a little worried when someone told me that the 270k pikel count was way low and I could get the same results with a 1ccd cam. This guy is telling me that i will only get about an 8% improvement With the gl-1 than a 1 chipper. Is there any truth behind this?

November 1st, 2002, 04:09 PM	#9
Jeff Donald Warden Join Date: Mar 2002 Location: Clearwater, FL Posts: 8,287	Everybody wants to narrow down performance to one number. Do you shop for cars by only looking at the Horse Power? Then don't shop for cameras by only looking at the number of pixels. Much more than total pixels goes into determining picture quality. Dre... and Don mention many of them above, lenses, diffraction, MTF, etc. Picture quality can not be determined by specs on a piece of paper. Try going to a retailer and looking at the images different cameras produce. Compare single chip camers to three chip cameras. I think you'll see more than a 8% difference. Jeff

November 1st, 2002, 04:52 PM	#10
Tylar Mccoy New Boot Join Date: Sep 2002 Posts: 18	Thanks, I had an optura100mc for about 2 weeks before I sent it back to onecall. (the 100mc also had a hot pixel) I decided to credit the 100mc and get a gl-1. When the gl-1 got here It too had a hot pixel and I only used it for about 20 minutes before packing it back up and sending it back. I have been looking at the gl-1 for about a year, and now that I can afford it im getting screwed around by a dealer. I cant wait untill I get a non defective unit so I can stop spending my time on this nightmare and just enjoy the product that I paid for over a month ago. I live in a rural area, So I would have to travel about 350 miles to find any stores high class enough to put 3 chippers on display, But for the short time I used the gl-1 I could notice a huge leap in quality. The most difference I noticed was shen I shot a blanket that is full of bright vivid colors. The same blanket shot with the 2 cams looks different for sure. I am a video newbie. Another thing I was reading is that the gl-1 is much worse in low light because of its 3ccds weakining the light. But I found the gl-1 worked better than the 100mc in low light, (in my living room filming my friends puppy)

November 2nd, 2002, 04:21 AM	#11
Andre De Clercq Major Player Join Date: Sep 2002 Location: Belgium Posts: 804	Where do you read an hear all those strange things, like the 8% "improvement", or the 3 CCD weakening the light performance? None of these statements is true.

November 2nd, 2002, 02:15 PM	#12
Tylar Mccoy New Boot Join Date: Sep 2002 Posts: 18	The 8% came from a guy who was reading one of my posts here, and emailed me. Heres part of his msg: "As for me, any 3CCD camcorders under $2400 are crap. Side by side comparison....single chipper and 3 chippers are 92/100%. Will you put out $800 extra to pay for 8% different? Picture quality is about 8% different as compared to 1chip JVC GR DV3000U $1000 range." The thing about the 3ccd's being worse in low light was something I read on a website, I cannot remember where, But it said something like, The Beam splits the light up, and as it splits the light onto the 3 different ccd's it get weakend, It said if you want good low light get a 1 chipper Because they are the most sensitive. Thanks for putting these myth's to rest for me.

November 2nd, 2002, 02:57 PM	#13
Jeff Donald Warden Join Date: Mar 2002 Location: Clearwater, FL Posts: 8,287	Single chip cameras generally perform better in low light than 3 chip cameras for the reason you mentioned. The beam splitter, splits the light for each chip, Red, Green, Blue. The 8% sounds like just one persons opinion to me. Jeff

November 2nd, 2002, 03:09 PM	#14
Andre De Clercq Major Player Join Date: Sep 2002 Location: Belgium Posts: 804	FYI camera's are not compaired in term of %. If all the parameters (sensitivity, resolution, lens qual;itiy...) would be the same (not realistic), single chippers allways show more aliasing effects (edge business, moire) and color fringeing. On the color splitting story, the guy posting this nonsence aparently doesn't know that a color filter in single CCD does the same in terms of allowing some colors only to some pixels. Worse even, a 3 CCD projects ALL the incoming light of a primary color on the related CCD, a single CCD filter allows 1/3 th of the incoming light (in case of a 3 color CCD filter) to specific pixels and absorbs the remaining light projected on those specific pixels. resulting in one of the lower sensitivity problems.