Alan Roberts test results available. at DVinfo.net

Steve Game · January 8th, 2012, 03:50 AM

For those who want to see an objective assessment of the C300, see here:

http://blog.creativevideo.co.uk/wp-c...an_roberts.pdf

Happy reading and real information.

Steve

Sanjin Svajger · January 8th, 2012, 08:16 AM

Quote:

Originally Posted by Alan Roberts

There are faint null zones due to
harmonic distortion, but the level is quite acceptable.

What is a "Null Zone"?

Alister Chapman · January 8th, 2012, 08:21 AM

Impressive results across the board. perhaps a tiny bit noisier at 0db than I was expecting, but no big deal. When you see how well the C300 performs with the XF305 10 bit processing and 8 bit output, it does make you wonder what the next Canon video camera with this sensor will be like. It should be even more impressive, possibly out perfuming and Alexa with the right recording format.

A null zone is where the output frequencies from the sensor or processing cancel each other out. The basic ring pattern of the zone plate explores the frequency response of the camera and is a very effective way to measure resolution as well as any issues with image detail beyond the cameras design resolution. A camera should be able to resolve detail (or frequencies) up to the sensors designed resolution. But if higher frequencies (higher resolution detail) than the sensor can correctly handle are allowed to fall on the sensor (or if there are gaps between pixels on the sensor), there will be some points where some details of the image falling on the sensor may exactly line up with the sensors pixels and this can create false detail. It's not true detail, but a coincidence where the rings on the chart may perfectly line up with pixels on the sensor. As the zone plate uses circles it explores every possible pattern to pixel alignment. This false detail will at certain frequencies match real detail and the two will either add together or subtract from each other and may cancel each other out and this creates the null zone, seen as a series of rings superimposed on the original rings of the chart.
A similar thing can happen if you take an image and electronically enhance it, as you are adding artificial detail. More specifically, a zone plate should create a sine wave pattern (pure undulating wave) if you looked at the signal on a scope. When you artificially sharpen the signal it may get turned into a square wave. A square wave contains many harmonic frequencies, which are multiples of the original frequency. These undesirable multiples will interact with the original pure sine wave frequency and again at certain frequencies cancel out the original signal and at others boost the original through subtraction and addition. This is why you see rings, superimposed within the original rings of the zone plate.

Steve Game · January 8th, 2012, 10:56 AM

To add to Alister's words on zone plates, their main use in camera testing is to verify the the sensor is not receiving detail that goes above the nyquist limit, i.e. half the sampling frequency of the sensor. To achieve this, the image coming through the lens must not be too sharp. Its normal with pro cameras to include an optical low pass filter to attenuate these higher frequency component sin the image.
As Alan seems to imply, unlike many other single chip cameras, Canon seem to have got this just right, with very little aliasing in the output signal. The biggest factor is that the C300 filter is set up for the camera's video resolution, i.e. 1929x1080, and there are no compromises that accommodate still photo images of much higher resolution. That compromise is what destroys the credibility of SLR video, forcing any serious user to avoid any scenes that have sharp or repetitive edges that generate an array of artifacts including moire.
Alan's is an interesting report based on real scientific measurement rather than reading specs., and then guessing.
I imagine that this camera will be BBC approved soon and available for use by those to whom shallow DoF is the most important thing in town. I only hope that any resulting TV work has other production qualities to be appreciated.

Steve

Robin Davies-Rollinson · January 8th, 2012, 01:52 PM

Quote:

Originally Posted by Steve Game

I only hope that any resulting TV work has other production qualities to be appreciated.

Steve

Well said, sir - I agree wholeheartedly!

Matt Ford · January 8th, 2012, 05:03 PM

Im confused!

Are you saying that drifting in and out of sharp focus won't be enough to win me a BAFTA?

Steve Game · January 8th, 2012, 05:35 PM

Well, you might need to sleep with the judges as well!

Steve

Alister Chapman · January 8th, 2012, 06:33 PM

Reading through the report again and comparing it with his earlier F3 report there are some differences in the way some of the data is presented that I think are a little confusing and could lead to some incorrect assumptions. In particular the noise measurements where for the C300 Alan quotes -54db (with noise reduction) and the F3 -48.5 db, but then the C300 was measured at -6db while the F3 at 0db. Using Alan's own plots the C300 at 0db is -45.5db (without noise reduction). So as expected a bit more noisy than the F3, not less noisy than the F3 as you might first assume from the way the report is written. I am also frustrated by the way in the C300 report the similar RGB zone plate alias results are is noted as a good thing while the similar RGB zone plate results for the F3 were noted as a bad thing.

Barry Goyette · January 8th, 2012, 09:14 PM

When I read Mr. Roberts tone in the noise section, he seems to be clearly evaluating the both cameras with respect to the manufacturers claimed S/N ratio. In canon's case he clearly points out several times that canon's stated noise floor is only achievable at -6db with noise reduction on, and with sony he's suggesting that they don't get even close at 0db gain to their stated 64db S/N ration. As the f3 can be used at -3db, it's possible that the camera could improve it's number slightly (or perhaps not), but certainly couldn't achieve it's claimed noise floor at negative gain.

In his comments, he gives both cameras really positive ratings regarding their noise levels.

I found the comments regarding aliasing with higher resolution lenses especially illuminating...

Barry

Murray Christian · January 9th, 2012, 12:55 AM

Quote:

Originally Posted by Steve Game

I imagine that this camera will be BBC approved soon and available for use by those to whom shallow DoF is the most important thing in town. I only hope that any resulting TV work has other production qualities to be appreciated.

This non sequitur seems like it's got some venom attached. What's the backstory there?

Steve Game · January 9th, 2012, 02:23 AM

Hardly venom. I have an ongoing observation that for many, shooting video with ridiculously shallow DoF is seen by some to qualify the work as cinematic and automatically on a par with the best of Hollywood. Focus control is just one of the cameraman's tools, not a style around which to model whole productions. I can't think of any acclaimed professional release that wallows in shallow DoF to the exclusion of all other cinematic techniques. It seems that at least one poster here agrees with me.

I suppose this is the wrong forum to comment on a fad that is held up by some as a sacred cow.

Steve

Sanjin Svajger · January 9th, 2012, 02:57 AM

Quote:

Originally Posted by Steve Game

I suppose this is the wrong forum to comment on a fad that is held up by some as a sacred cow.

Steve

Hardly. I think that most people here would agree with you.

Regarding the C300, as I'm reading through internet forums and interviews and now finally mr. Roberts reports it's interesting to see what this camera would have been capable of if they wouldn't rushed it and thus gave it a different hardware...

@Alister: thx for the answer.

Alister Chapman · January 9th, 2012, 03:01 AM

I too found that a little strange that the resolution and aliasing assessments were done with the zoom and the test results written with respect to the lower perfuming zoom, then subsequent tests found the primes to give better resolution with more aliasing. Assumptions were made about the OLPF and resolution from a lens that was clearly not performing as well as might be desirable. I'm not saying the results are wrong or incorrect, just I would have expected the resolution and aliasing testing to have been done with the best lens, not the worst lens. Clearly some tweaking of the detail settings will be needed to control the aliasing from higher resolution lenses.

According to his charts, the F3 at -3db achieved a -54db Y noise figure in Alan's test which is a fair bit less noise than the -48.5db of the C300 at -6db (without heavy noise reduction). I can't help but feel that Alan likes Canon cameras as he also comments on how the noise looks very filmic, which is unusual for one of his reports which are normally purely based on test results rather than subjective opinion.

Brian Drysdale · January 9th, 2012, 03:12 AM

Perhaps a case of not so much rushing as providing the market with a camera now, rather in maybe a year's time. Looking at other camera manufacturers that time scale seems to be what's involved in developing a full new camera from scratch after the sensor is available. I Imagine Canon talked to people to discover their reaction to the current C300 and had positive feedback, so put it into production. After a conversation with one of their HD experts, I got the impression that the BBC may have been across developments, even though he wasn't saying much at the time.

If you're making a film a camera now is much more useful.

Murray Christian · January 9th, 2012, 04:53 AM

Quote:

Originally Posted by Steve Game

Hardly venom. I have an ongoing observation that for many, shooting video with ridiculously shallow DoF is seen by some to qualify the work as cinematic and automatically on a par with the best of Hollywood. Focus control is just one of the cameraman's tools, not a style around which to model whole productions. I can't think of any acclaimed professional release that wallows in shallow DoF to the exclusion of all other cinematic techniques. It seems that at least one poster here agrees with me.

I suppose this is the wrong forum to comment on a fad that is held up by some as a sacred cow.

Steve

Never seen it sacred. Lots of people do like it though. There's probably a production or two that use it almost completely (to the exclusion of everything else I couldn't say. Presumably they would still have actors). A couple of Steven Soderberg's works spring to mind. But I haven't seen them in a while.
Anyway there's some people to whom it's a marker of quality. But c'est la vie. Mostly I think what you're seeing is cameras that can do it are falling into the hands of a lot of people who've been winging it on autofocus or whatever for a long time, because that's all they could afford/all the school had until very recently.
You made it sound as though the BBC had some weird policy of lots of shallow focus at the moment and the c300 would make it worse. That was all.

January 8th, 2012, 03:50 AM	#1
Steve Game Major Player Join Date: Mar 2011 Location: London, UK Posts: 353	Alan Roberts test results available. For those who want to see an objective assessment of the C300, see here: http://blog.creativevideo.co.uk/wp-c...an_roberts.pdf Happy reading and real information. Steve

January 8th, 2012, 08:21 AM	#3
Alister Chapman Inner Circle Join Date: Dec 2005 Location: Bracknell, Berkshire, UK Posts: 4,957	Re: Alan Roberts test results available. Impressive results across the board. perhaps a tiny bit noisier at 0db than I was expecting, but no big deal. When you see how well the C300 performs with the XF305 10 bit processing and 8 bit output, it does make you wonder what the next Canon video camera with this sensor will be like. It should be even more impressive, possibly out perfuming and Alexa with the right recording format. A null zone is where the output frequencies from the sensor or processing cancel each other out. The basic ring pattern of the zone plate explores the frequency response of the camera and is a very effective way to measure resolution as well as any issues with image detail beyond the cameras design resolution. A camera should be able to resolve detail (or frequencies) up to the sensors designed resolution. But if higher frequencies (higher resolution detail) than the sensor can correctly handle are allowed to fall on the sensor (or if there are gaps between pixels on the sensor), there will be some points where some details of the image falling on the sensor may exactly line up with the sensors pixels and this can create false detail. It's not true detail, but a coincidence where the rings on the chart may perfectly line up with pixels on the sensor. As the zone plate uses circles it explores every possible pattern to pixel alignment. This false detail will at certain frequencies match real detail and the two will either add together or subtract from each other and may cancel each other out and this creates the null zone, seen as a series of rings superimposed on the original rings of the chart. A similar thing can happen if you take an image and electronically enhance it, as you are adding artificial detail. More specifically, a zone plate should create a sine wave pattern (pure undulating wave) if you looked at the signal on a scope. When you artificially sharpen the signal it may get turned into a square wave. A square wave contains many harmonic frequencies, which are multiples of the original frequency. These undesirable multiples will interact with the original pure sine wave frequency and again at certain frequencies cancel out the original signal and at others boost the original through subtraction and addition. This is why you see rings, superimposed within the original rings of the zone plate. __________________ Alister Chapman, Film-Maker/Stormchaser http://www.xdcam-user.com/alisters-blog/ My XDCAM site and blog. http://www.hurricane-rig.com

January 8th, 2012, 10:56 AM	#4
Steve Game Major Player Join Date: Mar 2011 Location: London, UK Posts: 353	Re: Alan Roberts test results available. To add to Alister's words on zone plates, their main use in camera testing is to verify the the sensor is not receiving detail that goes above the nyquist limit, i.e. half the sampling frequency of the sensor. To achieve this, the image coming through the lens must not be too sharp. Its normal with pro cameras to include an optical low pass filter to attenuate these higher frequency component sin the image. As Alan seems to imply, unlike many other single chip cameras, Canon seem to have got this just right, with very little aliasing in the output signal. The biggest factor is that the C300 filter is set up for the camera's video resolution, i.e. 1929x1080, and there are no compromises that accommodate still photo images of much higher resolution. That compromise is what destroys the credibility of SLR video, forcing any serious user to avoid any scenes that have sharp or repetitive edges that generate an array of artifacts including moire. Alan's is an interesting report based on real scientific measurement rather than reading specs., and then guessing. I imagine that this camera will be BBC approved soon and available for use by those to whom shallow DoF is the most important thing in town. I only hope that any resulting TV work has other production qualities to be appreciated. Steve

January 8th, 2012, 05:03 PM	#6
Matt Ford Regular Crew Join Date: Dec 2010 Location: London Posts: 53	Re: Alan Roberts test results available. Im confused! Are you saying that drifting in and out of sharp focus won't be enough to win me a BAFTA?

January 8th, 2012, 05:35 PM	#7
Steve Game Major Player Join Date: Mar 2011 Location: London, UK Posts: 353	Re: Alan Roberts test results available. Well, you might need to sleep with the judges as well! Steve

January 8th, 2012, 06:33 PM	#8
Alister Chapman Inner Circle Join Date: Dec 2005 Location: Bracknell, Berkshire, UK Posts: 4,957	Re: Alan Roberts test results available. Reading through the report again and comparing it with his earlier F3 report there are some differences in the way some of the data is presented that I think are a little confusing and could lead to some incorrect assumptions. In particular the noise measurements where for the C300 Alan quotes -54db (with noise reduction) and the F3 -48.5 db, but then the C300 was measured at -6db while the F3 at 0db. Using Alan's own plots the C300 at 0db is -45.5db (without noise reduction). So as expected a bit more noisy than the F3, not less noisy than the F3 as you might first assume from the way the report is written. I am also frustrated by the way in the C300 report the similar RGB zone plate alias results are is noted as a good thing while the similar RGB zone plate results for the F3 were noted as a bad thing. __________________ Alister Chapman, Film-Maker/Stormchaser http://www.xdcam-user.com/alisters-blog/ My XDCAM site and blog. http://www.hurricane-rig.com Last edited by Alister Chapman; January 9th, 2012 at 03:14 AM.

January 8th, 2012, 09:14 PM	#9
Barry Goyette Trustee Join Date: Oct 2001 Location: San Luis Obispo CA Posts: 1,195	Re: Alan Roberts test results available. When I read Mr. Roberts tone in the noise section, he seems to be clearly evaluating the both cameras with respect to the manufacturers claimed S/N ratio. In canon's case he clearly points out several times that canon's stated noise floor is only achievable at -6db with noise reduction on, and with sony he's suggesting that they don't get even close at 0db gain to their stated 64db S/N ration. As the f3 can be used at -3db, it's possible that the camera could improve it's number slightly (or perhaps not), but certainly couldn't achieve it's claimed noise floor at negative gain. In his comments, he gives both cameras really positive ratings regarding their noise levels. I found the comments regarding aliasing with higher resolution lenses especially illuminating... Barry

January 9th, 2012, 02:23 AM	#11
Steve Game Major Player Join Date: Mar 2011 Location: London, UK Posts: 353	Re: Alan Roberts test results available. Hardly venom. I have an ongoing observation that for many, shooting video with ridiculously shallow DoF is seen by some to qualify the work as cinematic and automatically on a par with the best of Hollywood. Focus control is just one of the cameraman's tools, not a style around which to model whole productions. I can't think of any acclaimed professional release that wallows in shallow DoF to the exclusion of all other cinematic techniques. It seems that at least one poster here agrees with me. I suppose this is the wrong forum to comment on a fad that is held up by some as a sacred cow. Steve

January 9th, 2012, 03:01 AM	#13
Alister Chapman Inner Circle Join Date: Dec 2005 Location: Bracknell, Berkshire, UK Posts: 4,957	Re: Alan Roberts test results available. I too found that a little strange that the resolution and aliasing assessments were done with the zoom and the test results written with respect to the lower perfuming zoom, then subsequent tests found the primes to give better resolution with more aliasing. Assumptions were made about the OLPF and resolution from a lens that was clearly not performing as well as might be desirable. I'm not saying the results are wrong or incorrect, just I would have expected the resolution and aliasing testing to have been done with the best lens, not the worst lens. Clearly some tweaking of the detail settings will be needed to control the aliasing from higher resolution lenses. According to his charts, the F3 at -3db achieved a -54db Y noise figure in Alan's test which is a fair bit less noise than the -48.5db of the C300 at -6db (without heavy noise reduction). I can't help but feel that Alan likes Canon cameras as he also comments on how the noise looks very filmic, which is unusual for one of his reports which are normally purely based on test results rather than subjective opinion. __________________ Alister Chapman, Film-Maker/Stormchaser http://www.xdcam-user.com/alisters-blog/ My XDCAM site and blog. http://www.hurricane-rig.com

January 9th, 2012, 03:12 AM	#14
Brian Drysdale Inner Circle Join Date: Feb 2006 Location: Belfast, UK Posts: 6,152	Re: Alan Roberts test results available. Perhaps a case of not so much rushing as providing the market with a camera now, rather in maybe a year's time. Looking at other camera manufacturers that time scale seems to be what's involved in developing a full new camera from scratch after the sensor is available. I Imagine Canon talked to people to discover their reaction to the current C300 and had positive feedback, so put it into production. After a conversation with one of their HD experts, I got the impression that the BBC may have been across developments, even though he wasn't saying much at the time. If you're making a film a camera now is much more useful.