HD10 preliminary tests report at DVinfo.net

Eric Bilodeau · August 19th, 2003, 11:54 AM

I have been asked to do extensive tests on the HD10 camera in order to be able to shoot a feature with it and transfer to 35mm. I have tested it for 3 intensive days and I have shot a music video with it on sunday. Here are my preliminary findings:

The camera displays a slightly desaturated image (wich is fine by me as I always desaturate a little in post) but not too much, in good lighting condition it is almost not apparent.

This camera works really well in low light, surprisingly well (I will post some footage later this week). Black level seem to be much better that on DV (wich probably has to do with the compression method because DV has a tendency to loose a bit of the black). Usually, contrasts of the scene are kept accurately without too much "blast noise" (or if you prefer that silly dark edge that appears on the border of washed out reflections from shiny objects or from glowy objects themselves when the ccd overloads).

The downside however is a tendency to lose saturation in dark areas of the image mostly in sunlight situation or very bright lighting when some areas are not sufficiently bright or in shadows. I am now pretty sure that the loss in chroma comes from the exposure control itself (not the iris or the shutter); it seems as though the exposure dial, when set to -10 tends to lose chrominance partially or totally when light is low or is forced low by the manual settings. This might have to do with a boost of the luminance information witch takes over the chrominance in order to maintain data rate because the mpeg2 compression is a data stream and probably also to minimise compression noise.

The compression artefacts come from mostly the reds and blues, no artefacting seem to be created in the motion (wich is pretty impressive considering that the compression is based on a 6 pictures GOP, meaning one referrential image and the changes on the others). Also the artefacts from the chip are not as apparent as I would have guessed from a 1 chip CCD (if you control the exposure).

When you go from off to on in manual mode you lose the exposure lock but not the settings so you have to re-lock it.

The camera has a tendency to blow out whites, it is generally better to underexpose a little.

Most of the time, the iris will not go below f2 and the shutter will not go below 1/30. However, if you use manual control of the iris to darken the image the shutter gets down to 1/15th and you end up having a strobe effect. It is also true that opening the iris (to lessen depth of field for exemple) will end up boosting the shutter in a non desirable speed. It is best to trick the shutter and iris into a 1/30-1/60, f2 automatic adjustment, lock the exposure and use ND filters in order to maximise its performances.

I would say that the result has an average amount of compression artefact pretty close to DV. Head to head (I tested on an XL1s with manual lenses) they portray pretty much the same artefacts in the chrominance but the JVC seems to have less when loosing saturation.

Overall the major problem is still the compression artefacts as it is with DV. But as people seem to get over it (in DV), I think it is not such a big deal for most people.

The handling of the camera suits me, but handling is different from one person to another so I will not get onto that.

The image stabiliser works as well as the DV stabilisers I experienced from Canon and Sony.

In conclusion, I would recommend that camera to people used to DV but as it was said over and over again in these threads, you need an open minded approach as it lacks some pro features namely the individual iris and shutter manual controls. It is most definitely one of the best 1CCD I have worked with. It produces outstanding images when controlled.

More to come in the next weeks,

Eric

Paul Mogg · August 19th, 2003, 12:40 PM

Eric,
Great asessment, I agree with what you found pretty much 100% Though I haven't done any thorough low-light shooting tests myself yet, the couple of snippets I tried in low light seemed to come out suprisingly well.
Are you going to be doing any blow up to 35mm tests?, or digital screening tests?. I'd be very interested to know how the image from this camera converts to 35mm and, more interestingly, how it looks when projected Digitally on a very large (40') screen.
Just a reminder to people for the umpteenth time, NOT to judge the color noise or compression artifacts from this camera by viewing footage through a software playback program like Elecard or VLAN, as they create color noise and artifacts during playback, that are NOT in the original footage, the footage needs to be viewed directly from the camera (or from a DVHS deck) on a calibrated, native 720p capable HD monitor (not up or down converted), to judge quality issues. I promise not to repeat this again here!
I am still stunned by the cinematic nature of the 30p images that this camera can achieve under the right circumstances.

Cheers

Joe Russ · August 19th, 2003, 01:39 PM

<<<-- Originally posted by Paul Mogg : Eric,
Great asessment, I agree with what you found pretty much 100% Though I haven't done any thorough low-light shooting tests myself yet, the couple of snippets I tried in low light seemed to come out suprisingly well.
Are you going to be doing any blow up to 35mm tests?, or digital screening tests?. I'd be very interested to know how the image from this camera converts to 35mm and, more interestingly, how it looks when projected Digitally on a very large (40') screen.
Just a reminder to people for the umpteenth time, NOT to judge the color noise or compression artifacts from this camera by viewing footage through a software playback program like Elecard or VLAN, as they create color noise and artifacts during playback, that are NOT in the original footage, the footage needs to be viewed directly from the camera (or from a DVHS deck) on a calibrated, native 720p capable HD monitor (not up or down converted), to judge quality issues. I promise not to repeat this again here!
I am still stunned by the cinematic nature of the 30p images that this camera can achieve under the right circumstances.

Cheers -->>>

actually, the color noise is still there, these media players arent helping the video quality but they arent making up noise that isnt there. i converted the footage to uncompressed using a round about method (not using videolan or elecard or a dvhs deck) and as soon as you adjust the levels you can see the chroma noise, its just the price of such a low bitrate for compression, they preserved the luma at cost of the chroma (a wise choice). also, id have to say low light is pretty bad on this camera, all the 3 ccd dv cameras i have used have better low light capabilities....but then again, im not using it in those kinda situations, so it shouldnt matter all that much....maybe for documentary stuff.

if you want to see what the chroma noise looks like (not coming from those software players), go here:

http://www.letterstoyou.com/noise00.jpg

in the test you can clearly see the chroma noise, i showed ther chroma noise after i set the levels (which i have to do anyway and just brings out the noise more) but other then that its untouched.

Paul Mogg · August 19th, 2003, 02:34 PM

Well, all I can tell you is that I've done quite a few tests comparing the same footage played directly from the camera to viewing it through the Elecard and VLAN players on the same monitor, and through this I've found that these software players most definately DO add significant color noise that is not there in the original footage. and accentuate noise that is there.
This is most easily visible in cloudless blue skies, if you'd like to test it for yourself. This suprised me quite a bit, but it is most definately the case.
This does not mean that there is NO color noise in the original footage, there is, visible especially in more saturated reds, but if you want to make a fair judgement as to how severe it is, you should definately NOT do so through the software players.
These are just my personal findings and opinions that I've come to through my own experimentation, and using a Monivision 32" monitor in native 720p mode to view the footage on.

Frank Granovski · August 19th, 2003, 03:14 PM

Eric, what transfer lab are you going to use? I don't know of any transfer labs that recommend or are even geared up to use the new MPEG2 HD format. Are you going to be using another mode for the transfer? All I'm saying is that check with the transfer lab you will be using first to see what they require, along with their do's and do not's.

Eric Bilodeau · August 19th, 2003, 03:35 PM

transfers will depend on a lot of factors, mainly the ability to convert to 24p either in some software like after effects plugins or directly where the film will be blown. 24p in software transfer seems really good. I talked with Cam Morrisson from JVC Canada and they had a hunch I had myself (wich could or could not work): upconvert (to DVHS) in 1080i and from the "false 60i" footage created, transfer to 35mm. I have doubts but who knows. If it does not work, we will have to transfer 24p via a software and put this 24p footage at a speed of 30p (boosting the speed significantly) and there could then be a transfer frame by frame. The other solution is to perform a pulldown (either standard or the new 3:4:4:3 supported by both the DVX100 and FCP4) to DVHS 1080i and that's it. Anyway the chain is not completely figured out yet.

One thing is sure for the moment, we plan to transfer at Digital Film Group in Vancouver.

As for the noise and VideoLan or a HD monitor we have to be careful on that. VLC does not create that much noise in the decompression, the noise I was talking abbout was directly visible on the monitor. But a software view of the material is most of the time worst than an analog view (like in a HD or SD monitor). The worst artefacts I have seen so far where on a plasma screen.
I noticed more than once a very beautiful image of some DV footage on some monitors that I knew whithout a doubt had originally noise... Remember that NTSC is "Never Twice the Same Color", that goes for HD monitors as well from what I have seen so far. It depends on a lot of factors. I agree though that I see more noise on VideoLan than on the original footage, that doesn't meen noise was originally absent.

Steve Mullen · August 19th, 2003, 05:11 PM

<<<-- Originally posted by Eric Bilodeau : I talked with Cam Morrisson from JVC Canada and they had a hunch I had myself (which could or could not work): upconvert (to DVHS) in 1080i and from the "false 60i" footage created, transfer to 35mm. I have doubts, but who knows.
-->>>

You won't change the temporal sampling frequency doing this so when you do the pull-down to 24P from 60i -- you are still converting to 24p from 30p.

<<<-- The other solution is to perform a pulldown (either standard or the new 3:4:4:3 supported by both the DVX100 and FCP4) to DVHS 1080i and that's it. -->>>

That's not the 24p full-down -- but it doesn't matter. It's scanned at 30fps at the CCD.

<<<-- The worst artifacts I have seen so far where on a plasma screen. -->>>

That's because most plasmas are crap. Plan to spend > $6K. More likely, $10K!

GREAT REVIEW Eric!

You confirmed all my findings which is always a good thing! And you are right about red and blue chroma noise.

Ken Hodson · August 24th, 2003, 04:30 PM

"The compression artefacts come from mostly the reds and blues, no artefacting seem to be created in the motion (wich is pretty impressive considering that the compression is based on a 6 pictures GOP, meaning one referrential image and the changes on the others). Also the artefacts from the chip are not as apparent as I would have guessed from a 1 chip CCD (if you control the exposure)."

Does this mean "Green Screen" removal ability would be decent then?
I am hoping that a "magic bullet" type deartifactor coupled with the 720p will give me the leap I need over what has become quite competent DV green screen removal.

Eric Bilodeau · August 24th, 2003, 07:09 PM

I believe that green screen removal might be a bit easier than DV because of two facts mostly: compression is present in the same proportion (or at least pretty closely) but definition is 2.6 times greater so it is likely that green screen will be easier because mostly of the leap in definition considering that artefacts seem to be in the same intensity. The more overall light, the less artefacts so lighting will be a major issue (as it is in DV). I should do some green screen tests in a few weeks, I will keep you informed.

Eric

Ken Hodson · August 25th, 2003, 12:27 PM

Please do.
Thanks Eric

August 19th, 2003, 11:54 AM	#1
Eric Bilodeau Major Player Join Date: Jun 2003 Location: Chicoutimi, Canada Posts: 334	HD10 preliminary tests report I have been asked to do extensive tests on the HD10 camera in order to be able to shoot a feature with it and transfer to 35mm. I have tested it for 3 intensive days and I have shot a music video with it on sunday. Here are my preliminary findings: The camera displays a slightly desaturated image (wich is fine by me as I always desaturate a little in post) but not too much, in good lighting condition it is almost not apparent. This camera works really well in low light, surprisingly well (I will post some footage later this week). Black level seem to be much better that on DV (wich probably has to do with the compression method because DV has a tendency to loose a bit of the black). Usually, contrasts of the scene are kept accurately without too much "blast noise" (or if you prefer that silly dark edge that appears on the border of washed out reflections from shiny objects or from glowy objects themselves when the ccd overloads). The downside however is a tendency to lose saturation in dark areas of the image mostly in sunlight situation or very bright lighting when some areas are not sufficiently bright or in shadows. I am now pretty sure that the loss in chroma comes from the exposure control itself (not the iris or the shutter); it seems as though the exposure dial, when set to -10 tends to lose chrominance partially or totally when light is low or is forced low by the manual settings. This might have to do with a boost of the luminance information witch takes over the chrominance in order to maintain data rate because the mpeg2 compression is a data stream and probably also to minimise compression noise. The compression artefacts come from mostly the reds and blues, no artefacting seem to be created in the motion (wich is pretty impressive considering that the compression is based on a 6 pictures GOP, meaning one referrential image and the changes on the others). Also the artefacts from the chip are not as apparent as I would have guessed from a 1 chip CCD (if you control the exposure). When you go from off to on in manual mode you lose the exposure lock but not the settings so you have to re-lock it. The camera has a tendency to blow out whites, it is generally better to underexpose a little. Most of the time, the iris will not go below f2 and the shutter will not go below 1/30. However, if you use manual control of the iris to darken the image the shutter gets down to 1/15th and you end up having a strobe effect. It is also true that opening the iris (to lessen depth of field for exemple) will end up boosting the shutter in a non desirable speed. It is best to trick the shutter and iris into a 1/30-1/60, f2 automatic adjustment, lock the exposure and use ND filters in order to maximise its performances. I would say that the result has an average amount of compression artefact pretty close to DV. Head to head (I tested on an XL1s with manual lenses) they portray pretty much the same artefacts in the chrominance but the JVC seems to have less when loosing saturation. Overall the major problem is still the compression artefacts as it is with DV. But as people seem to get over it (in DV), I think it is not such a big deal for most people. The handling of the camera suits me, but handling is different from one person to another so I will not get onto that. The image stabiliser works as well as the DV stabilisers I experienced from Canon and Sony. In conclusion, I would recommend that camera to people used to DV but as it was said over and over again in these threads, you need an open minded approach as it lacks some pro features namely the individual iris and shutter manual controls. It is most definitely one of the best 1CCD I have worked with. It produces outstanding images when controlled. More to come in the next weeks, Eric __________________ Eric Bilodeau video SFX,DOP ___________________ http://www.fictis.net info@fictis.net

August 19th, 2003, 12:40 PM	#2
Paul Mogg Major Player Join Date: Apr 2003 Location: San Francisco CA Posts: 386	Eric, Great asessment, I agree with what you found pretty much 100% Though I haven't done any thorough low-light shooting tests myself yet, the couple of snippets I tried in low light seemed to come out suprisingly well. Are you going to be doing any blow up to 35mm tests?, or digital screening tests?. I'd be very interested to know how the image from this camera converts to 35mm and, more interestingly, how it looks when projected Digitally on a very large (40') screen. Just a reminder to people for the umpteenth time, NOT to judge the color noise or compression artifacts from this camera by viewing footage through a software playback program like Elecard or VLAN, as they create color noise and artifacts during playback, that are NOT in the original footage, the footage needs to be viewed directly from the camera (or from a DVHS deck) on a calibrated, native 720p capable HD monitor (not up or down converted), to judge quality issues. I promise not to repeat this again here! I am still stunned by the cinematic nature of the 30p images that this camera can achieve under the right circumstances. Cheers __________________ Paul

August 19th, 2003, 02:34 PM	#4
Paul Mogg Major Player Join Date: Apr 2003 Location: San Francisco CA Posts: 386	Well, all I can tell you is that I've done quite a few tests comparing the same footage played directly from the camera to viewing it through the Elecard and VLAN players on the same monitor, and through this I've found that these software players most definately DO add significant color noise that is not there in the original footage. and accentuate noise that is there. This is most easily visible in cloudless blue skies, if you'd like to test it for yourself. This suprised me quite a bit, but it is most definately the case. This does not mean that there is NO color noise in the original footage, there is, visible especially in more saturated reds, but if you want to make a fair judgement as to how severe it is, you should definately NOT do so through the software players. These are just my personal findings and opinions that I've come to through my own experimentation, and using a Monivision 32" monitor in native 720p mode to view the footage on. __________________ Paul

August 19th, 2003, 03:35 PM	#6
Eric Bilodeau Major Player Join Date: Jun 2003 Location: Chicoutimi, Canada Posts: 334	transfers will depend on a lot of factors, mainly the ability to convert to 24p either in some software like after effects plugins or directly where the film will be blown. 24p in software transfer seems really good. I talked with Cam Morrisson from JVC Canada and they had a hunch I had myself (wich could or could not work): upconvert (to DVHS) in 1080i and from the "false 60i" footage created, transfer to 35mm. I have doubts but who knows. If it does not work, we will have to transfer 24p via a software and put this 24p footage at a speed of 30p (boosting the speed significantly) and there could then be a transfer frame by frame. The other solution is to perform a pulldown (either standard or the new 3:4:4:3 supported by both the DVX100 and FCP4) to DVHS 1080i and that's it. Anyway the chain is not completely figured out yet. One thing is sure for the moment, we plan to transfer at Digital Film Group in Vancouver. As for the noise and VideoLan or a HD monitor we have to be careful on that. VLC does not create that much noise in the decompression, the noise I was talking abbout was directly visible on the monitor. But a software view of the material is most of the time worst than an analog view (like in a HD or SD monitor). The worst artefacts I have seen so far where on a plasma screen. I noticed more than once a very beautiful image of some DV footage on some monitors that I knew whithout a doubt had originally noise... Remember that NTSC is "Never Twice the Same Color", that goes for HD monitors as well from what I have seen so far. It depends on a lot of factors. I agree though that I see more noise on VideoLan than on the original footage, that doesn't meen noise was originally absent. __________________ Eric Bilodeau video SFX,DOP ___________________ http://www.fictis.net info@fictis.net

August 19th, 2003, 05:11 PM	#7
Steve Mullen HDV Cinema Join Date: Feb 2003 Location: Las Vegas Posts: 4,007	<<<-- Originally posted by Eric Bilodeau : I talked with Cam Morrisson from JVC Canada and they had a hunch I had myself (which could or could not work): upconvert (to DVHS) in 1080i and from the "false 60i" footage created, transfer to 35mm. I have doubts, but who knows. -->>> You won't change the temporal sampling frequency doing this so when you do the pull-down to 24P from 60i -- you are still converting to 24p from 30p. <<<-- The other solution is to perform a pulldown (either standard or the new 3:4:4:3 supported by both the DVX100 and FCP4) to DVHS 1080i and that's it. -->>> That's not the 24p full-down -- but it doesn't matter. It's scanned at 30fps at the CCD. <<<-- The worst artifacts I have seen so far where on a plasma screen. -->>> That's because most plasmas are crap. Plan to spend > $6K. More likely, $10K! GREAT REVIEW Eric! You confirmed all my findings which is always a good thing! And you are right about red and blue chroma noise. __________________ Switcher's Quick Guide to the Avid Media Composer >>> http://home.mindspring.com/~d-v-c

August 19th, 2003, 01:39 PM	#3
Joe Russ Regular Crew Join Date: Jul 2003 Location: New York, NY Posts: 99	<<<-- Originally posted by Paul Mogg : Eric, Great asessment, I agree with what you found pretty much 100% Though I haven't done any thorough low-light shooting tests myself yet, the couple of snippets I tried in low light seemed to come out suprisingly well. Are you going to be doing any blow up to 35mm tests?, or digital screening tests?. I'd be very interested to know how the image from this camera converts to 35mm and, more interestingly, how it looks when projected Digitally on a very large (40') screen. Just a reminder to people for the umpteenth time, NOT to judge the color noise or compression artifacts from this camera by viewing footage through a software playback program like Elecard or VLAN, as they create color noise and artifacts during playback, that are NOT in the original footage, the footage needs to be viewed directly from the camera (or from a DVHS deck) on a calibrated, native 720p capable HD monitor (not up or down converted), to judge quality issues. I promise not to repeat this again here! I am still stunned by the cinematic nature of the 30p images that this camera can achieve under the right circumstances. Cheers -->>> actually, the color noise is still there, these media players arent helping the video quality but they arent making up noise that isnt there. i converted the footage to uncompressed using a round about method (not using videolan or elecard or a dvhs deck) and as soon as you adjust the levels you can see the chroma noise, its just the price of such a low bitrate for compression, they preserved the luma at cost of the chroma (a wise choice). also, id have to say low light is pretty bad on this camera, all the 3 ccd dv cameras i have used have better low light capabilities....but then again, im not using it in those kinda situations, so it shouldnt matter all that much....maybe for documentary stuff. if you want to see what the chroma noise looks like (not coming from those software players), go here: http://www.letterstoyou.com/noise00.jpg in the test you can clearly see the chroma noise, i showed ther chroma noise after i set the levels (which i have to do anyway and just brings out the noise more) but other then that its untouched.

August 19th, 2003, 03:14 PM	#5
Frank Granovski Outer Circle Join Date: May 2002 Location: Hope, BC Posts: 7,524	Eric, what transfer lab are you going to use? I don't know of any transfer labs that recommend or are even geared up to use the new MPEG2 HD format. Are you going to be using another mode for the transfer? All I'm saying is that check with the transfer lab you will be using first to see what they require, along with their do's and do not's.

August 24th, 2003, 04:30 PM	#8
Ken Hodson Trustee Join Date: Aug 2003 Location: Vancouver BC Canada Posts: 1,315	"The compression artefacts come from mostly the reds and blues, no artefacting seem to be created in the motion (wich is pretty impressive considering that the compression is based on a 6 pictures GOP, meaning one referrential image and the changes on the others). Also the artefacts from the chip are not as apparent as I would have guessed from a 1 chip CCD (if you control the exposure)." Does this mean "Green Screen" removal ability would be decent then? I am hoping that a "magic bullet" type deartifactor coupled with the 720p will give me the leap I need over what has become quite competent DV green screen removal. __________________ Damnit Jim, I'm a film maker not a sysytems tech.

August 24th, 2003, 07:09 PM	#9
Eric Bilodeau Major Player Join Date: Jun 2003 Location: Chicoutimi, Canada Posts: 334	I believe that green screen removal might be a bit easier than DV because of two facts mostly: compression is present in the same proportion (or at least pretty closely) but definition is 2.6 times greater so it is likely that green screen will be easier because mostly of the leap in definition considering that artefacts seem to be in the same intensity. The more overall light, the less artefacts so lighting will be a major issue (as it is in DV). I should do some green screen tests in a few weeks, I will keep you informed. Eric __________________ Eric Bilodeau video SFX,DOP ___________________ http://www.fictis.net info@fictis.net

August 25th, 2003, 12:27 PM	#10
Ken Hodson Trustee Join Date: Aug 2003 Location: Vancouver BC Canada Posts: 1,315	Please do. Thanks Eric __________________ Damnit Jim, I'm a film maker not a sysytems tech.