Topics

Camera evaluations - request for comments

Geoff Boyle
 

First, my apologies for cross posting, but if I can’t break my own rules…I’m also posting on Cinerant.

 

The next camera evaluations are shooting in three weeks and I’ve been thinking about how to change them from previous ones, bearing in mind comments I’ve received.

 

I’m going to try to move away from marking =/- on the charts, it will be purely stop and shutter angle for this test.

 

I will light to 200fc and set the cameras to the manufacturers recommended EI.

 

We will shoot from T1.5 to T 22 at 180 degrees and then T22 at 90 & 45 degree shutter.

 

The tests will be shot both tungsten and LED daylight. Tungsten being the most “accurate” source I can use. I’d love to take the tests to Australia to shoot across a window of 11 am to 1pm across the weeks it would take to shoot the tests in real daylight! The LED’s will be the most commonly used and should show what happens in a “real-world” environment.

 

The evaluations will be available to download as EXR’s in ACES space from the ODN, and hopefully, depending on cost and support, I’ll also make some of the original camera files available. I had intended to do this but we lost a number of sponsors this year and may not be able to afford it.

 

I will also upload QT files to Vimeo, these will be UHD and 422, I’ll let them convert them to H264, and also H264 directly to YouTube.

 

The compiled files that I upload will be all matched at 18% grey, matching only the green level as I don’t intend to colour correct them in any way. Everything will go through ACES with standard IDT’s and appropriate ODT’s.

 

This time I will make observations as to which EI I think the cameras should be exposed at to get the best images, this will be basically based around noise levels.

So if you set the camera to an EI of 800 but the noise is better at 500 I’ll say so, I’ll also upload a LUT that adjust for you setting the camera to 800 and your meter to 500. This will be a LUT in 2 forms, one that is “naked” and just needs to be dropped into the ACES pipeline in Resolve and the other will include the appropriate IDT’s and ODT’s to “correct” the image on 709 monitors on the floor.

 

I guess I’m going back to the days when I published GRR’s for various film stocks 😊

 

OK, beat me up now.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

Derek Abel
 

Thanks for doing this Geoff.

Don’t come to Australia for ‘real daylight’ just yet. The weather here at the moment is almost as bad as the UK

:-)


Derek Abel
Filmmaker/ Digital Content Creator (in Australia)
0449534995

www.DerekAbel.com
www.FocalRhythm.com
www.ZoeShortFilm.com
www.HomelessAustralia.com

On 17 Jun 2018, at 18:08, Geoff Boyle <geoff@...> wrote:

I’d love to take the tests to Australia to shoot across a window of 11 am to 1pm across the weeks it would take to shoot the tests in real daylight!

Mitch Gross
 

A few questions to which I suspect you mostly have answers but did not indicate. 


What will you point the cameras at?

You did not say so, but will you shoot in RAW or Log if the camera offers it? If it offers both will you shoot both? If it does not offer either how will you shoot it , or does this disqualify the camera from the test?

How will you set white balance on the cameras, preset or auto off a grey card?

What are your restrictions for resolution? What about cameras with multiple resolution options, or cameras with a higher resolution sensor than the output format? Will all camera outputs be transformed to UHD no matter the starting resolution? I ask because you mentioned noise, which could be affected by this parameter. 

What about cameras that do not offer a direct solution for your planned workflow? Such as a ProRes RAW recording. 


Mitch Gross
Cinema Product Manager 
Panasonic Systems Solutions Company of North America
New York

On Jun 17, 2018, at 4:08 AM, Geoff Boyle <geoff@...> wrote:

First, my apologies for cross posting, but if I can’t break my own rules…I’m also posting on Cinerant.

 

The next camera evaluations are shooting in three weeks and I’ve been thinking about how to change them from previous ones, bearing in mind comments I’ve received.

 

I’m going to try to move away from marking =/- on the charts, it will be purely stop and shutter angle for this test.

 

I will light to 200fc and set the cameras to the manufacturers recommended EI.

 

We will shoot from T1.5 to T 22 at 180 degrees and then T22 at 90 & 45 degree shutter.

 

The tests will be shot both tungsten and LED daylight. Tungsten being the most “accurate” source I can use. I’d love to take the tests to Australia to shoot across a window of 11 am to 1pm across the weeks it would take to shoot the tests in real daylight! The LED’s will be the most commonly used and should show what happens in a “real-world” environment.

 

The evaluations will be available to download as EXR’s in ACES space from the ODN, and hopefully, depending on cost and support, I’ll also make some of the original camera files available. I had intended to do this but we lost a number of sponsors this year and may not be able to afford it.

 

I will also upload QT files to Vimeo, these will be UHD and 422, I’ll let them convert them to H264, and also H264 directly to YouTube.

 

The compiled files that I upload will be all matched at 18% grey, matching only the green level as I don’t intend to colour correct them in any way. Everything will go through ACES with standard IDT’s and appropriate ODT’s.

 

This time I will make observations as to which EI I think the cameras should be exposed at to get the best images, this will be basically based around noise levels.

So if you set the camera to an EI of 800 but the noise is better at 500 I’ll say so, I’ll also upload a LUT that adjust for you setting the camera to 800 and your meter to 500. This will be a LUT in 2 forms, one that is “naked” and just needs to be dropped into the ACES pipeline in Resolve and the other will include the appropriate IDT’s and ODT’s to “correct” the image on 709 monitors on the floor.

 

I guess I’m going back to the days when I published GRR’s for various film stocks 😊

 

OK, beat me up now.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

Geoff Boyle
 

Hmm, a lot of questions from Mitch and a ton more points for people to pick at me 😊

 

As before DSC charts plus a moving element.

 

RAW whenever possible, I think it is in all the cameras we are testing this time, UWE has a bunch of Shoguns so I suspect that ProRes RAW will get an outing.

 

No cameras are disqualified, it’s simple, the top of a manufacturers range and the lowest that I feel acceptable 😊

 

Preset white balance.

 

Cameras that are multi image size I will test at FF & S35, where they also down sample I will test that but obviously that can’t be RAW and will be log probably to the compressed on-board recorders.

 

I will aim to record at 4K and crop to UHD when creating deliverables, this is what I do in the real world and I see no reason to change. The 6K and above will obviously be rendered down rather than cropped.

 

I am optimistic about a solution for ProRes being available by then, if not I’ll maybe have to go CDNG instead. That’s up to Apple.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

From: cml-raw-log-hdr@... <cml-raw-log-hdr@...> On Behalf Of Mitch Gross
Sent: 17 June 2018 12:04
To: cml-raw-log-hdr@...
Subject: Re: [cml-raw-log-hdr] Camera evaluations - request for comments

 

A few questions to which I suspect you mostly have answers but did not indicate. 

 

 

What will you point the cameras at?

 

You did not say so, but will you shoot in RAW or Log if the camera offers it? If it offers both will you shoot both? If it does not offer either how will you shoot it , or does this disqualify the camera from the test?

 

How will you set white balance on the cameras, preset or auto off a grey card?

 

What are your restrictions for resolution? What about cameras with multiple resolution options, or cameras with a higher resolution sensor than the output format? Will all camera outputs be transformed to UHD no matter the starting resolution? I ask because you mentioned noise, which could be affected by this parameter. 

 

What about cameras that do not offer a direct solution for your planned workflow? Such as a ProRes RAW recording. 

 

 

Mitch Gross

Cinema Product Manager 

Panasonic Systems Solutions Company of North America

New York


On Jun 17, 2018, at 4:08 AM, Geoff Boyle <geoff@...> wrote:

First, my apologies for cross posting, but if I can’t break my own rules…I’m also posting on Cinerant.

 

The next camera evaluations are shooting in three weeks and I’ve been thinking about how to change them from previous ones, bearing in mind comments I’ve received.

 

I’m going to try to move away from marking =/- on the charts, it will be purely stop and shutter angle for this test.

 

I will light to 200fc and set the cameras to the manufacturers recommended EI.

 

We will shoot from T1.5 to T 22 at 180 degrees and then T22 at 90 & 45 degree shutter.

 

The tests will be shot both tungsten and LED daylight. Tungsten being the most “accurate” source I can use. I’d love to take the tests to Australia to shoot across a window of 11 am to 1pm across the weeks it would take to shoot the tests in real daylight! The LED’s will be the most commonly used and should show what happens in a “real-world” environment.

 

The evaluations will be available to download as EXR’s in ACES space from the ODN, and hopefully, depending on cost and support, I’ll also make some of the original camera files available. I had intended to do this but we lost a number of sponsors this year and may not be able to afford it.

 

I will also upload QT files to Vimeo, these will be UHD and 422, I’ll let them convert them to H264, and also H264 directly to YouTube.

 

The compiled files that I upload will be all matched at 18% grey, matching only the green level as I don’t intend to colour correct them in any way. Everything will go through ACES with standard IDT’s and appropriate ODT’s.

 

This time I will make observations as to which EI I think the cameras should be exposed at to get the best images, this will be basically based around noise levels.

So if you set the camera to an EI of 800 but the noise is better at 500 I’ll say so, I’ll also upload a LUT that adjust for you setting the camera to 800 and your meter to 500. This will be a LUT in 2 forms, one that is “naked” and just needs to be dropped into the ACES pipeline in Resolve and the other will include the appropriate IDT’s and ODT’s to “correct” the image on 709 monitors on the floor.

 

I guess I’m going back to the days when I published GRR’s for various film stocks 😊

 

OK, beat me up now.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

Geoff Boyle
 

OOPs!

 

400 FC and down to 22.5 degree shutter.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

From: cml-raw-log-hdr@... <cml-raw-log-hdr@...> On Behalf Of Geoff Boyle
Sent: 17 June 2018 10:08
To: cml-general@...; cml-raw-log-hdr@...
Subject: [cml-raw-log-hdr] Camera evaluations - request for comments

 

First, my apologies for cross posting, but if I can’t break my own rules…I’m also posting on Cinerant.

 

The next camera evaluations are shooting in three weeks and I’ve been thinking about how to change them from previous ones, bearing in mind comments I’ve received.

 

I’m going to try to move away from marking =/- on the charts, it will be purely stop and shutter angle for this test.

 

I will light to 200fc and set the cameras to the manufacturers recommended EI.

 

We will shoot from T1.5 to T 22 at 180 degrees and then T22 at 90 & 45 degree shutter.

 

The tests will be shot both tungsten and LED daylight. Tungsten being the most “accurate” source I can use. I’d love to take the tests to Australia to shoot across a window of 11 am to 1pm across the weeks it would take to shoot the tests in real daylight! The LED’s will be the most commonly used and should show what happens in a “real-world” environment.

 

The evaluations will be available to download as EXR’s in ACES space from the ODN, and hopefully, depending on cost and support, I’ll also make some of the original camera files available. I had intended to do this but we lost a number of sponsors this year and may not be able to afford it.

 

I will also upload QT files to Vimeo, these will be UHD and 422, I’ll let them convert them to H264, and also H264 directly to YouTube.

 

The compiled files that I upload will be all matched at 18% grey, matching only the green level as I don’t intend to colour correct them in any way. Everything will go through ACES with standard IDT’s and appropriate ODT’s.

 

This time I will make observations as to which EI I think the cameras should be exposed at to get the best images, this will be basically based around noise levels.

So if you set the camera to an EI of 800 but the noise is better at 500 I’ll say so, I’ll also upload a LUT that adjust for you setting the camera to 800 and your meter to 500. This will be a LUT in 2 forms, one that is “naked” and just needs to be dropped into the ACES pipeline in Resolve and the other will include the appropriate IDT’s and ODT’s to “correct” the image on 709 monitors on the floor.

 

I guess I’m going back to the days when I published GRR’s for various film stocks 😊

 

OK, beat me up now.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

Art Adams
 

A suggestion about charts:

I'd skip the Chroma Match. It's not useful for this kind of test. I'd like to see Chroma Du Monde and your custom chart, or even just a One Shot and your custom chart.

I'd also include a Color Checker, although I think you usually do. The DSC Labs charts (excluding your custom chart) are built primarily for use with a waveform, and since you can't make a chart where a color channel dips to zero, each color has a percentage of the others. (For example, the green chip viewed in parade RGB should show the green channel at 80% and the others at 40%.) This is an amazing way to judge color crossover between channels, which is valuable, but only on a waveform. It also allows one to set up a camera using a waveform alone, which is also valuable. Visually, though, these hues can be a little deceptive as none of them are pure, and as few of us will be looking at those chart in that way I'd suggest having both the DSC charts plus something that has known pure colors.

I'd love to see you use shutter as the primary means of exposure control so changes in lens contrast with stop changes don't come into play. I know that's likely impossible across the range of stops that you're going to test, though.

I would like to see the cameras white balanced. Preset is, in my experience, not a great way to work. Some cameras have great presets, others are greenish. I don't think seeing the differences in the camera presets is useful because those are most likely to change in the field or in post anyway. Also, color shifts will interfere with our ability to compare color rendition between cameras.

The idea of matching exposures using the green channel is interesting but potentially problematic. As the luma channel is comprised of roughly 70% green, 20% and 10% blue (not exact numbers) I'd think your overall luma would end up being too bright by about 30% if exposure is based only on the green channel. If you don't white balance and the preset is a little skewed then apparently brightness levels between cameras will vary based on that. Matching the exposures based on the luma channel after white balancing would probably be the most accurate.

I'd also like to see the middle gray exposure based on a code value that is commonly accepted as middle gray, if such a thing exists. 512 is generally too bright, and that will skew noise results.

Looking forward to the results...

--
Art Adams
Director of Photography
San Francisco Bay Area

Geoff Boyle
 

I’m still thinking about charts, my one shot is too battered to use now. I do have standard X-Rite charts and the VFX DSC charts.

 

As for auto V preset, I keep changing my mind 😊 on the one hand I auto balance when I start a scene so… on the other hand the manufacturers should get it right and so I should show when they don’t.

 

I think my explanation of only using the green channel is somewhat misunderstood. I mean in this context that if the camera is off-colour then when I use offset to adjust the level I alter the offset level so that the green signal is on the line that I am aiming for. There’s not a lot in it but when there is a colour variation I need to have a standard way of working.

 

As for the levels, we’ll never agree on this 😊 for the people that give a shit there will be EXR files and OCN. For those that watch on YouTube and Vimeo I set the levels that experience has taught me works best there. I always make a point of mentioning this when I publish the results.

 

Where is middle grey 😊 in ACES 0.18.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

From: cml-raw-log-hdr@... <cml-raw-log-hdr@...> On Behalf Of Art Adams
Sent: 17 June 2018 18:35
To: Cml-Raw-Log-Hdr <cml-raw-log-hdr@...>
Subject: Re: [cml-raw-log-hdr] Camera evaluations - request for comments

 

A suggestion about charts:

 

I'd skip the Chroma Match. It's not useful for this kind of test. I'd like to see Chroma Du Monde and your custom chart, or even just a One Shot and your custom chart.

 

I'd also include a Color Checker, although I think you usually do. The DSC Labs charts (excluding your custom chart) are built primarily for use with a waveform, and since you can't make a chart where a color channel dips to zero, each color has a percentage of the others. (For example, the green chip viewed in parade RGB should show the green channel at 80% and the others at 40%.) This is an amazing way to judge color crossover between channels, which is valuable, but only on a waveform. It also allows one to set up a camera using a waveform alone, which is also valuable. Visually, though, these hues can be a little deceptive as none of them are pure, and as few of us will be looking at those chart in that way I'd suggest having both the DSC charts plus something that has known pure colors.

 

I'd love to see you use shutter as the primary means of exposure control so changes in lens contrast with stop changes don't come into play. I know that's likely impossible across the range of stops that you're going to test, though.

 

I would like to see the cameras white balanced. Preset is, in my experience, not a great way to work. Some cameras have great presets, others are greenish. I don't think seeing the differences in the camera presets is useful because those are most likely to change in the field or in post anyway. Also, color shifts will interfere with our ability to compare color rendition between cameras.

 

The idea of matching exposures using the green channel is interesting but potentially problematic. As the luma channel is comprised of roughly 70% green, 20% and 10% blue (not exact numbers) I'd think your overall luma would end up being too bright by about 30% if exposure is based only on the green channel. If you don't white balance and the preset is a little skewed then apparently brightness levels between cameras will vary based on that. Matching the exposures based on the luma channel after white balancing would probably be the most accurate.

 

I'd also like to see the middle gray exposure based on a code value that is commonly accepted as middle gray, if such a thing exists. 512 is generally too bright, and that will skew noise results.

 

Looking forward to the results...

 

--

Art Adams

Director of Photography

San Francisco Bay Area

 

Mitch Gross
 

I vote for having the cameras pull a white balance off a grey card. Let’s forget about presets and what some engineers thought might happen in the real world and instead look at what the camera thinks it is seeing. I think this is the more pure and useful way. 

Mitch Gross
Cinema Product Manager 
Panasonic Systems Solutions Company of North America
New York

On Jun 17, 2018, at 2:18 PM, Geoff Boyle <geoff.cml@...> wrote:

I’m still thinking about charts, my one shot is too battered to use now. I do have standard X-Rite charts and the VFX DSC charts.

 

As for auto V preset, I keep changing my mind 😊 on the one hand I auto balance when I start a scene so… on the other hand the manufacturers should get it right and so I should show when they don’t.

 

I think my explanation of only using the green channel is somewhat misunderstood. I mean in this context that if the camera is off-colour then when I use offset to adjust the level I alter the offset level so that the green signal is on the line that I am aiming for. There’s not a lot in it but when there is a colour variation I need to have a standard way of working.

 

As for the levels, we’ll never agree on this 😊 for the people that give a shit there will be EXR files and OCN. For those that watch on YouTube and Vimeo I set the levels that experience has taught me works best there. I always make a point of mentioning this when I publish the results.

 

Where is middle grey 😊 in ACES 0.18.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Netherlands

www.gboyle.nl

 

From: cml-raw-log-hdr@... <cml-raw-log-hdr@...> On Behalf Of Art Adams
Sent: 17 June 2018 18:35
To: Cml-Raw-Log-Hdr <cml-raw-log-hdr@...>
Subject: Re: [cml-raw-log-hdr] Camera evaluations - request for comments

 

A suggestion about charts:

 

I'd skip the Chroma Match. It's not useful for this kind of test. I'd like to see Chroma Du Monde and your custom chart, or even just a One Shot and your custom chart.

 

I'd also include a Color Checker, although I think you usually do. The DSC Labs charts (excluding your custom chart) are built primarily for use with a waveform, and since you can't make a chart where a color channel dips to zero, each color has a percentage of the others. (For example, the green chip viewed in parade RGB should show the green channel at 80% and the others at 40%.) This is an amazing way to judge color crossover between channels, which is valuable, but only on a waveform. It also allows one to set up a camera using a waveform alone, which is also valuable. Visually, though, these hues can be a little deceptive as none of them are pure, and as few of us will be looking at those chart in that way I'd suggest having both the DSC charts plus something that has known pure colors.

 

I'd love to see you use shutter as the primary means of exposure control so changes in lens contrast with stop changes don't come into play. I know that's likely impossible across the range of stops that you're going to test, though.

 

I would like to see the cameras white balanced. Preset is, in my experience, not a great way to work. Some cameras have great presets, others are greenish. I don't think seeing the differences in the camera presets is useful because those are most likely to change in the field or in post anyway. Also, color shifts will interfere with our ability to compare color rendition between cameras.

 

The idea of matching exposures using the green channel is interesting but potentially problematic. As the luma channel is comprised of roughly 70% green, 20% and 10% blue (not exact numbers) I'd think your overall luma would end up being too bright by about 30% if exposure is based only on the green channel. If you don't white balance and the preset is a little skewed then apparently brightness levels between cameras will vary based on that. Matching the exposures based on the luma channel after white balancing would probably be the most accurate.

 

I'd also like to see the middle gray exposure based on a code value that is commonly accepted as middle gray, if such a thing exists. 512 is generally too bright, and that will skew noise results.

 

Looking forward to the results...

 

--

Art Adams

Director of Photography

San Francisco Bay Area

 

Mark Weingartner, ASC
 




On 17Jun, 2018, at 16:43 21, Mitch Gross <mitchgrosscml@...> wrote:

I vote for having the cameras pull a white balance off a grey card. Let’s forget about presets and what some engineers thought might happen in the real world and instead look at what the camera thinks it is seeing. I think this is the more pure and useful way. 

Funny - I feel just the opposite:   Since I know I can null out differences between the mfr’s belief of what a neutral color balance is and what shows up, I would rather see the cameras at their pre-sets… look at what the mfr considered to be optimal for the two color temperatures.

At least that way you are avoiding one more procedure that might vary slightly from one camera’s white balance to the next…           

I admit to my bias - whether motion picture or still camera I very rarely auto-white unless I am in a situation with uncontrollable odd-spectrum lighting - a supermarket lit by cool white fluorescents, for instance

I am much happier going with a pre-set and retiming as necessary in post so that if using multiple cameras or shooting in the same location or stage on multiple days I don’t introduce slight differences in footage from day to day which might slow down coloring the edited scene.

Some of this may come from my laziness at keeping at it the way one would with film - the native color balance of the stock is a “preset” - because in my HL-79 days we certainly white balanced a lot…
…but then again tape-to-tape color correction was less sophisticated then and if you could get it right in camera it was less likely to get cocked up by someone in post.

My two cents

Mark Weingartner, ASC
DP inclined towards Visual Effects
Los Angeles-based

Art Adams
 

Shooting on preset means extra work white balancing in post later. There'll need to be tweaks even with white balancing in set, but at least they'd be close.

Not white balancing the cameras is a great way to see, as Mitch said, what some engineer in a lab thought was a good preset. Without white balancing the cameras at some point, whether in production or post, critical color analysis becomes impossible. At that point it's just a contest between engineers and their presets.

--
Art Adams
DP
San Francisco Bay Area



On Jun 17, 2018 at 5:45 PM, <ASC Mark Weingartner> wrote:




On 17Jun, 2018, at 16:43 21, Mitch Gross <mitchgrosscml@...> wrote:

I vote for having the cameras pull a white balance off a grey card. Let’s forget about presets and what some engineers thought might happen in the real world and instead look at what the camera thinks it is seeing. I think this is the more pure and useful way. 

Funny - I feel just the opposite:   Since I know I can null out differences between the mfr’s belief of what a neutral color balance is and what shows up, I would rather see the cameras at their pre-sets… look at what the mfr considered to be optimal for the two color temperatures.

At least that way you are avoiding one more procedure that might vary slightly from one camera’s white balance to the next…           

I admit to my bias - whether motion picture or still camera I very rarely auto-white unless I am in a situation with uncontrollable odd-spectrum lighting - a supermarket lit by cool white fluorescents, for instance

I am much happier going with a pre-set and retiming as necessary in post so that if using multiple cameras or shooting in the same location or stage on multiple days I don’t introduce slight differences in footage from day to day which might slow down coloring the edited scene.

Some of this may come from my laziness at keeping at it the way one would with film - the native color balance of the stock is a “preset” - because in my HL-79 days we certainly white balanced a lot…
…but then again tape-to-tape color correction was less sophisticated then and if you could get it right in camera it was less likely to get cocked up by someone in post.

My two cents

Mark Weingartner, ASC
DP inclined towards Visual Effects
Los Angeles-based

Mark Weingartner, ASC
 


On 17Jun, 2018, at 18:01 47, Art Adams <art.cml.only@...> wrote:

Without white balancing the cameras at some point, whether in production or post, critical color analysis becomes impossible. At that point it's just a contest between engineers and their presets.

Not to quibble but critical color analysis is certainly possible without either white balancing before shooting or white balancing in post - you may not like the camera’s rendition but if you have shot standard referents you can look at the offsets between the recorded image and the known values of those referents without re-balancing anything.

You might not want to deliver a project you shoot that way, but you can certainly analyze it.

Different DPs and different types of jobs suggest different ways of working - if we are doing on-set color management under “normal” illuminants the people I work with are more likely to be starting from pre-sets than from an auto-white balance situation.  Since I mostly work on features and someone else is shooting first unit, whether I tweak the white balance to nudge the images where I want them to go versus sticking to an agreed-upon CT setting vs using low saturation LB filters on the camera (as I did on a biggish feature a few years ago) is usually the topic of a prior discussion.

For me, pre-sets are a much better place to start in normal circumstances, so for me being able to compare how different cameras render a common scene under presets gives me insight into what I have to think about if I am shooting with a different type of camera from main unit (which happens all the time in my world.)

Mark Weingartner, ASC
LA - based DP inclined towards VFX


Geoff Boyle
 

My tendency is to agree with Mark, except thst in the dim and distant days of bvp300's i didn't white balance, i used presets and got compliments from editors about how well my footage worked. Of course i owned the cameras and we made sure that 3200 was 3200 with no shifts.
An engineers opinion of a preset??? WTF?
Sorry guys but 3200 is 3200 with zero offset in the M/G. Its not an opinion its a fact.
And this thread has confirmed my belief thst preset is the way to go. If the camera engineers cant manage to make a neutral tungsten then maybe we should take that as a warning!
Daylight LED's are a different matter and maybe white balancing them is a good idea because it shows how well the cameras cope with questionable sources.

cheers
Geoff Boyle NSC
EU based cinematographer
+31 637155076
www.gboyle.co.uk


On Mon, 18 Jun 2018, 03:13 Mark Weingartner, ASC, <vfxmark@...> wrote:

On 17Jun, 2018, at 18:01 47, Art Adams <art.cml.only@...> wrote:

Without white balancing the cameras at some point, whether in production or post, critical color analysis becomes impossible. At that point it's just a contest between engineers and their presets.

Not to quibble but critical color analysis is certainly possible without either white balancing before shooting or white balancing in post - you may not like the camera’s rendition but if you have shot standard referents you can look at the offsets between the recorded image and the known values of those referents without re-balancing anything.

You might not want to deliver a project you shoot that way, but you can certainly analyze it.

Different DPs and different types of jobs suggest different ways of working - if we are doing on-set color management under “normal” illuminants the people I work with are more likely to be starting from pre-sets than from an auto-white balance situation.  Since I mostly work on features and someone else is shooting first unit, whether I tweak the white balance to nudge the images where I want them to go versus sticking to an agreed-upon CT setting vs using low saturation LB filters on the camera (as I did on a biggish feature a few years ago) is usually the topic of a prior discussion.

For me, pre-sets are a much better place to start in normal circumstances, so for me being able to compare how different cameras render a common scene under presets gives me insight into what I have to think about if I am shooting with a different type of camera from main unit (which happens all the time in my world.)

Mark Weingartner, ASC
LA - based DP inclined towards VFX


Mark Kenfield
 

I'll add a +1 for going by presets.

Are there really that many people out there who set their white balances differently? I'd be very surprised if there are (percentage-wise).

If the ambient light in a scene is a bit off (and I don't want the white balance to be off), I'll take a reading with my spectrometer and set my camera to that reading - but it's still always the camera's built-in preset for that particular colour temperature, I never have the time on set to run around and frame up grey card shots illuminated by the ambient sources, and then run the camera through a white balance calibration process - it's way too time-consuming. 

If I'm particularly worried (or if I have multiple camera to match up in post), I'll throw a grey card in with the slate so that we have a reference point. But that's about as much of a concession as I can afford to make for it.

Cheers,

Mark Kenfield
Cinematographer

0400 044 500

On 18 June 2018 at 12:54, Geoff Boyle <geoff@...> wrote:
My tendency is to agree with Mark, except thst in the dim and distant days of bvp300's i didn't white balance, i used presets and got compliments from editors about how well my footage worked. Of course i owned the cameras and we made sure that 3200 was 3200 with no shifts.
An engineers opinion of a preset??? WTF?
Sorry guys but 3200 is 3200 with zero offset in the M/G. Its not an opinion its a fact.
And this thread has confirmed my belief thst preset is the way to go. If the camera engineers cant manage to make a neutral tungsten then maybe we should take that as a warning!
Daylight LED's are a different matter and maybe white balancing them is a good idea because it shows how well the cameras cope with questionable sources.

cheers
Geoff Boyle NSC
EU based cinematographer
+31 637155076
www.gboyle.co.uk

On Mon, 18 Jun 2018, 03:13 Mark Weingartner, ASC, <vfxmark@...> wrote:

On 17Jun, 2018, at 18:01 47, Art Adams <art.cml.only@...> wrote:

Without white balancing the cameras at some point, whether in production or post, critical color analysis becomes impossible. At that point it's just a contest between engineers and their presets.

Not to quibble but critical color analysis is certainly possible without either white balancing before shooting or white balancing in post - you may not like the camera’s rendition but if you have shot standard referents you can look at the offsets between the recorded image and the known values of those referents without re-balancing anything.

You might not want to deliver a project you shoot that way, but you can certainly analyze it.

Different DPs and different types of jobs suggest different ways of working - if we are doing on-set color management under “normal” illuminants the people I work with are more likely to be starting from pre-sets than from an auto-white balance situation.  Since I mostly work on features and someone else is shooting first unit, whether I tweak the white balance to nudge the images where I want them to go versus sticking to an agreed-upon CT setting vs using low saturation LB filters on the camera (as I did on a biggish feature a few years ago) is usually the topic of a prior discussion.

For me, pre-sets are a much better place to start in normal circumstances, so for me being able to compare how different cameras render a common scene under presets gives me insight into what I have to think about if I am shooting with a different type of camera from main unit (which happens all the time in my world.)

Mark Weingartner, ASC
LA - based DP inclined towards VFX


Mark Weingartner, ASC
 


On 17Jun, 2018, at 19:54 10, Geoff Boyle <geoff@...> wrote:

Daylight LED's are a different matter and maybe white balancing them is a good idea because it shows how well the cameras cope with questionable sources.

Hmmm interesting “compromise”     
Certainly logic to it  - preset for the established standard and “custom” auto-white balance for the discontinuous spectrum prevailing illumination for the daylight.

That slightly mitigates the aspect of the test that is investigating the concatenation of the spectral power curve of the lights and the spectral sensitivity curves of the various cameras.

Weingartner
la-based DP vfx etc



Ted Hayash
 

First things first: thank you Geoff for the most reliable and comprehensive tests available outside of those I do myself. It’s one thing to do tests for oneself, or for a specific production or purpose, it’s a whole other thing to produce a set of tests that will stand up to the scrutiny of the world, and Geoff’s tests do just that.

My vote would be to use a preset white balance as some of the cameras only have presets (Sony Venice comes to mind, although I think that’s about to change with the advent of new firmware), and although it’s an extra step in post, it’s a much larger extra step during the test. At least if I were doing the test, I’d use a card or chart that filled the frame with a specific light source, so that each camera is white balanced in a consistent manner. I think I’d rather skip that extra set of variables, as they create more risk of a mistake during the test. The preset is a better control than my process. After having done many tests myself over the years, and even been paid to do them, eliminating variables is of paramount importance, including limiting the potential for human fatigue and error.

I also tend to use presets in studio or on location when the light is consistent. It’s a standard against which to judge the balance of the lighting in the frame, and is more consistent over a number of days as well. I can count on the preset to be consistent - even if it’s one that I create myself through testing. I won’t generally auto white balance unless there are mixed or unusual sources in the scene, and even then I will use the auto white balance as a starting point.

Ted Hayash
Cinematographer
Los Angeles

 Ted Hayash
Ted Hayash
Director of Photography 
www.tedhayash.com


On June 18, 2018 at 1:13 GMT, Mark Weingartner ASC <vfxmark@...> wrote:


On 17Jun, 2018, at 18:01 47, Art Adams <art.cml.only@...> wrote:

Without white balancing the cameras at some point, whether in production or post, critical color analysis becomes impossible. At that point it's just a contest between engineers and their presets.

Not to quibble but critical color analysis is certainly possible without either white balancing before shooting or white balancing in post - you may not like the camera’s rendition but if you have shot standard referents you can look at the offsets between the recorded image and the known values of those referents without re-balancing anything.

You might not want to deliver a project you shoot that way, but you can certainly analyze it.

Different DPs and different types of jobs suggest different ways of working - if we are doing on-set color management under “normal” illuminants the people I work with are more likely to be starting from pre-sets than from an auto-white balance situation.  Since I mostly work on features and someone else is shooting first unit, whether I tweak the white balance to nudge the images where I want them to go versus sticking to an agreed-upon CT setting vs using low saturation LB filters on the camera (as I did on a biggish feature a few years ago) is usually the topic of a prior discussion.

For me, pre-sets are a much better place to start in normal circumstances, so for me being able to compare how different cameras render a common scene under presets gives me insight into what I have to think about if I am shooting with a different type of camera from main unit (which happens all the time in my world.)

Mark Weingartner, ASC
LA - based DP inclined towards VFX



Mitch Gross
 

I thought you were testing the recording capabilities of various cameras. To me that means neutralizing other factors. Now you are also comparing white balance presets. 

Mitch Gross
Cinema Product Manager 
Panasonic Systems Solutions Company of North America
New York


Geoff Boyle
 

Hi Mitch,

I don't see it like that, I'm comparing the cameras and trying to remove the recording capabilities from the equation by using the best possible.
As 3200 Kelvin is an absolute i don't see anything wrong with checking if the cameras match that absolute value.
3200 Kelvin isn't something approximate or open to discussion. To me it's an absolute fundamental.
We get to play with how well the cameras can auto white balance when we come to daylight.

cheers
Geoff Boyle NSC
EU based cinematographer
+31 637155076
www.gboyle.co.uk


On Mon, 18 Jun 2018, 14:52 Mitch Gross, <mitchgrosscml@...> wrote:
I thought you were testing the recording capabilities of various cameras. To me that means neutralizing other factors. Now you are also comparing white balance presets. 

Mitch Gross
Cinema Product Manager 
Panasonic Systems Solutions Company of North America
New York


Art Adams
 

>Not to quibble but critical color analysis is certainly possible without either white balancing before shooting or white balancing in post - you may not like the camera’s rendition but if you have shot standard referents you can look at the offsets between the recorded image and the known values of those referents without re-balancing anything.

No, it's not.

Sure, I could view the footage from one camera wearing rose tinted glasses and footage from another wearing green tinted glasses, but I won't be able to compare their color rendition at all. To say that this is possible is... I don't know why you'd say this is possible.

Some camera presets are fine. Some aren't. Some we get used to. Arri presets are known to be a little green, which shows up typically as a bit of extra yellow in the image. Is it a big deal? Maybe not normally, but when comparing images with other cameras, like the F55, where the Cine-EI presets are dead on, it'll make a difference.

Then there's Canon, whose standard reds have a greenish (yellowish) hue and greens are very cool. If the white balance isn't dead on then you're not going to being able to know what's a white balance error and what's a manufacturer's chosen look.

Varicams tend to punch yellows in a way that other cameras don't—that is to say they are accurate, and don't tend to skew green easily, which is a big problem with most cameras. Sky blues are stunning. If the white balance isn't neutral, though, you won't be able to appreciate these subtleties.

The neutrals must be matched at some point. Otherwise the tests are pointless. I don't understand why there is any argument about this at all. Shoot them in preset, fine. But match the middle gray values in post at the very least.

--
Art Adams
Director of Photography
San Francisco Bay Area

Art Adams
 

As 3200 Kelvin is an absolute

This target is an absolute. Nothing about color is an absolute. No two cameras are going to render this in exactly the same way, as cameras are not absolute machines when it comes to color reproduction. They can't be. All you can compare is how they do it differently across models and brands.

--
Art Adams
Director of Photography
San Francisco Bay Area

Geoff Boyle
 

How often do i have to say thst the tests are to show the fucking differences not to hide them!

cheers
Geoff Boyle NSC
EU based cinematographer
+31 637155076
www.gboyle.co.uk

On Mon, 18 Jun 2018, 16:02 Art Adams, <art.cml.only@...> wrote:
>Not to quibble but critical color analysis is certainly possible without either white balancing before shooting or white balancing in post - you may not like the camera’s rendition but if you have shot standard referents you can look at the offsets between the recorded image and the known values of those referents without re-balancing anything.

No, it's not.

Sure, I could view the footage from one camera wearing rose tinted glasses and footage from another wearing green tinted glasses, but I won't be able to compare their color rendition at all. To say that this is possible is... I don't know why you'd say this is possible.

Some camera presets are fine. Some aren't. Some we get used to. Arri presets are known to be a little green, which shows up typically as a bit of extra yellow in the image. Is it a big deal? Maybe not normally, but when comparing images with other cameras, like the F55, where the Cine-EI presets are dead on, it'll make a difference.

Then there's Canon, whose standard reds have a greenish (yellowish) hue and greens are very cool. If the white balance isn't dead on then you're not going to being able to know what's a white balance error and what's a manufacturer's chosen look.

Varicams tend to punch yellows in a way that other cameras don't—that is to say they are accurate, and don't tend to skew green easily, which is a big problem with most cameras. Sky blues are stunning. If the white balance isn't neutral, though, you won't be able to appreciate these subtleties.

The neutrals must be matched at some point. Otherwise the tests are pointless. I don't understand why there is any argument about this at all. Shoot them in preset, fine. But match the middle gray values in post at the very least.

--
Art Adams
Director of Photography
San Francisco Bay Area