Web Analytics
   Date   

Camera evaluations

Geoff Boyle
 

I’m not going to identify cameras but here are the five main manufacturers top cameras as of June 2017…

 

The values are RGB and were read with the picker tool in Resolve in ACEScct using the SDK or IDT as appropriate.

 

75         77         73

126       119       124

135       135       135

138       126       130

147       148       143

 

An older camera that I hate but has been popular 50       56              50

 

Venice   141       133       139

 

My only comment is that all cameras were set to pre-set daylight and ha the same lighting setup with the exception of the Venice which had a different lighting setup which may account for the colour difference, the light level was the same.

 

Cheers

 

Geoff Boyle NSC FBKS

Cinematographer

Zoetermeer

www.gboyle.co.uk

+31 (0) 637 155 076

 


Re: Venice evaluation

Geoff Boyle
 

Oh yes, a very clear idea but I'm not getting involved in that shitstorm!

The truth is out there...EXRs

Cheers
Geoff Boyle NSC
Cinematographer
EU based

Sent from Blue

On 27 Mar 2018, at 04:47, Art Adams <art.cml.only@...> wrote:
>Yes there are variations between cameras. FMR that's what this is all about.

Any idea what the differences were compared to a calibrated meter?

--
Art Adams
Director of Photography
San Francisco Bay Area


Re: Venice evaluation

Art Adams <art.cml.only@...>
 

>Yes there are variations between cameras. FMR that's what this is all about.

Any idea what the differences were compared to a calibrated meter?

--
Art Adams
Director of Photography
San Francisco Bay Area


Re: Venice evaluation

Geoff Boyle
 

What do I have to do to get it through to you?

The EXRs are generated with the manufacturer's software, if there are variations in level between cameras this is because, err, there are variations between cameras.

It's not a workflow issue! It's reality, it's what the damn things produce.

Yes there are variations between cameras. FMR that's what this is all about.

Cheers
Geoff Boyle NSC
Cinematographer
EU based

Sent from Blue

On 27 Mar 2018, at 01:18, "Michael Nagel via Cml.News" <yahoo.com@cml.news target=_blank>michanagel=yahoo.com@cml.news> wrote:
I've not looked at the SXT frames, but if they are that far off, then yeah workflow would have to be addressed.

Maybe if you propose how to level mid-grey he will adjust that in the next round ?

But whether you "shoot a scene" or charts with peeps (caucasian/black) and a flower in the shot, which btw is also a scene, both will still give you the color/texture change over these exposure ranges. At least that's how I understood was one of the main points of these tests. (please correct if wrong)

In a landscape scene though, the gradient of the sky (and how it breaks apart) and foliage would be another strong indicator of how these cams perform in real-world under/over shots...

- Mike Nagel
Director/Producer
L.A.


Re: Venice evaluation

Geoff Boyle
 

Thank you, you get it.

Cheers
Geoff Boyle NSC
Cinematographer
EU based

Sent from Blue

On 27 Mar 2018, at 00:56, "Michael Nagel via Cml.News" <yahoo.com@cml.news target=_blank>michanagel=yahoo.com@cml.news> wrote:
50% IRE for mid-grey used to be the "standard reference".

there's nothing here that "works" because this is a non-scientific test. He's just leveling the playing field by doing the same thing for all cams involved. (at least this is how I understand it)

The other thing is, since you don't know everybody's viewing environment, where would you place mid-grey ? How would you know whether a 2.2 or 2.4 gamma assessment is "correct"... ? I've calibrated screens for peeps with a 2.6 gamma, worked just fine for them and their environment... you don't whether the user is viewing on CRT/LCD/LED/OLED/RPTV or projectors...

Also, the mid-grey chip gain adjustment he does is just to ballpark the actual EI/ISO of the cam. Again, non-scientific. This is just to see whether there is a strong deviation from what the manufacturer claims. At least that's how I understand his workflow.

the over/under frames are to see how color/texture holds up and when/how noise gets introduced. IMO, this is kinda drastic on some cams, see also the 2015 tests. Now, obviously you can address some of that in post, but that was not the point of the comparison.

all of the over/under have the same reference point (0 exposure), whether you think that's "too bright" or not doesn't really matter - this is all relative - since nobody knows the true EI of the cam and the manufacturer's claims can NEVER be trusted. ;-)

- Mike Nagel
Director/Producer
L.A.


Re: Venice evaluation

Geoff Boyle
 

I've already said that if the absolute accuracy is what you want then the EXRs are there.

I've been doing the QTs like this for a very long time.

There are examples on CML from nearly 20 years ago, if I was to change you'd be screaming that I wasn't being consistent and giving me shit for it.

If you grade by your numbers Art then the white levels are way down and the shadows are compressed.

Now this may be "correct" but it looks crap.

The way I do the QTs reflects a real world use of the images.

I do the grade on an Eizo 318 in Rec 709, it should be sRGB but I want the file that you can download to look right on a 709 monitor, please don't argue with my choice of right.
I don't deliver clients a technical grade, I deli something that looks right.

I check the Vimeo files on  a PC using Chrome & Firefox on a Dell 2414 and on a MacBook Pro using Chrome & Safari.

Once again, and finally as I'm not wasting any more time on this, the EXRs are there if you want technical totallity, the Vimeo files are there if you want a real world comparison.
The YouTube files are there if you make assessments on your phone!

Cheers
Geoff Boyle NSC
Cinematographer
EU based

Sent from Blue

On 27 Mar 2018, at 00:26, Art Adams <art.cml.only@...> wrote:
Yeah, but there's always some sort of gamma applied, which means middle gray is not the same as splitting the code value range in half—not in Rec 709, or P3, or any log curve, or linear... not in anything.

You *can* put 18% gray wherever you want, but if you don't reproduce it in a way that resembles 18% gray on a display or screen of some sort, it's not 18% gray. The whole point is that it is a standard reference. If you don't use it that way, it loses its value. For example, when the ID on the EXR frames say "+5" or "-2"... in relation to what, if middle gray isn't middle gray? And isn't this deceptive if one just wants to look at the 0% frame and see how it grades? Doesn't it affect one's perception of noise in all the other frames?

And isn't judging how a camera looks all about perception?

If middle gray ever came to 50 IRE then those days are long gone by over a decade.

So sure, you *can* grade all this footage so middle gray is perceptually brighter than it should be, but then why include exposure information? It doesn't really relate to anything other than "plus or minus f/stops in relation to how someone thinks this footage looks good on Vimeo." That's not really a standard, and makes it a little more difficult to interpret the results across other display methods.

(I'm usually in the minority when it comes to this stuff, so if this works for you, cool. For me, it isn't a clean way to examine a camera.)

--
Art Adams
Director of Photography
San Francisco Bay Area


Re: New Red sensor.

Philip Holland
 

RED Gemini.
https://www.youtube.com/watch?v=xcKPupoLtKM

This is an entirely new sensor design with dual sensitivity.


Phil


-----------------
Phil Holland - Cinematographer
http://www.phfx.com
http://www.imdb.com/name/nm0390802/
818 470 0623


From: cml-raw-log-hdr@... <cml-raw-log-hdr@...> on behalf of dave@... <dave@...>
Sent: Monday, March 26, 2018 3:54:02 PM
To: cml-raw-log-hdr@...
Subject: Re: [raw-log-hdr] New Red sensor.
 
The link doesn't go to the video.

David Brillhart
Cinematographer
Sacramento, CA


Re: Clients and monitors (was Venice evaluation)

David Fuller
 

I’ve done that as well. Not fun. And then graded it back when he looked at it on his office computer the next morning. 

I’ve taken to asking clients to look at work on an iPad, because at this point it seems like the most reliable screen an art director is likely to have.

David Fuller
Director, Cinematographer
Maine, USA

---

On Mar 26, 2018, at 3:42 PM, Gavin Greenwalt <im.thatoneguy@...> wrote:

The real problem is that no matter how well educated you become or how well calibrated you are, in my experience all that matters is how it looks on the random corporate issued laptop that the Creative Director reviews it on.   We literally had to regrade a piece once because a VP of marketing saw it on air on his home TV and decided the blacks were too dark!


Re: New Red sensor.

David Brillhart
 

The link doesn't go to the video.

David Brillhart
Cinematographer
Sacramento, CA


locked Re: Venice evaluation

David Fuller
 

Art, isn’t this just semantics? By that I mean that ASA and EI refer to the same thing—or at least as I understand them they do—only ASA is the manufacturer’s recommendation and EI is “what i set my meter to.” In the film days, I often had a different EI from the manufacturer’s ASA rating. Is that different from what you’re saying or is there more to it that I’m not understanding?


David Fuller
Director, Cinematographer, Camera Geek
Maine, USA

---

On Mar 26, 2018, at 12:41 PM, Art Adams <art.cml.only@...> wrote:

I do wish we'd get away from using ISO and switch to EI, the way Arri uses it. EI means "Set your meter to this and you'll get predictable results." ISO is a standard, but in theory there's only one standard per camera, so as soon as you set a different "ISO" on the camera the standard is out the window. And the standard is so malleable—or so I've been told—that it's effectively meaningless.



locked Re: Venice evaluation

 

Thanks Steve.. That answered my question on Post-VFX :-)

Michael Sanders
London Based DP.

+ 44 (0) 7976 269818




On 26 Mar 2018, at 19:06, Steve Shaw <steve@...> wrote:

We do not recommend BT1886 for display calibration, as the 'variable'
component is ill thought-out.
It will 'wash-out' shadows on displays with blacks that are not close to
zero.


locked Re: Venice evaluation

Bruce Alan Greene
 

Hi Steve, I have a kind of non-technical observation.

When viewing Arri LogC through the Arri-REC709 LUT, it seems to me that it “fits” better into gamma 2.2 than gamma 2.4  

I’m not sure how one can really measure this in a scientific way, but I find that working in gamma 2.4 using this transform, I’m often using a gamma adjustment after the LUT conversion to 2.4, but not if my display is at 2.2…

What do you think Steve?  Are the Arri LogC to REC709 LUTs really built for gamma 2.2?  Is there really anyway to know this as camera exposure really effects the result as well…?

Bruce Alan Greene
DP Los Angeles


On Mar 26, 2018, at 11:06 AM, Steve Shaw <steve@...> wrote:

We do not recommend BT1886 for display calibration, as the 'variable'
component is ill thought-out.
It will 'wash-out' shadows on displays with blacks that are not close to
zero.

We recommend all calibration to be standardised at 2.4 gamma for Rec709, as
that is the 'underlying' target of BT1886, and is now also targeted for
Rec709.
sRGB is a 2.2 gamma for display calibration - the 'compound' encoding gamma
is for, well, encoding (capture).

Gamma 2.4 is standard for Reference Viewing Environments, with a surround
illumination of 10% of peak display luma (so 10 nits, as 100 nits is the
peak luma for Rec709).
(SMTE have tried to alter this to 5 nits regardless, but that is excessively
low, and is based on the needs of PQ based HDR, where the EOTF crushed
blacks causing them to be wiped-out in brighter environments.)

In Home Viewing Environments a lower gamma can be used (2.2 for example) to
attempt to overcome the brighter viewing environments, but should not be
relied on for any colour critical environments.

For any colour critical viewing you must manage the environment, not alter
the gamma.

Steve

Steve Shaw
LIGHT ILLUSION
steve@...
+44 (0)7765 400 908
www.lightillusion.com


Re: Venice evaluation

Mike Nagel
 

I've not looked at the SXT frames, but if they are that far off, then yeah workflow would have to be addressed.

Maybe if you propose how to level mid-grey he will adjust that in the next round ?

But whether you "shoot a scene" or charts with peeps (caucasian/black) and a flower in the shot, which btw is also a scene, both will still give you the color/texture change over these exposure ranges. At least that's how I understood was one of the main points of these tests. (please correct if wrong)

In a landscape scene though, the gradient of the sky (and how it breaks apart) and foliage would be another strong indicator of how these cams perform in real-world under/over shots...

- Mike Nagel
Director/Producer
L.A.


Re: Venice evaluation

Art Adams <art.cml.only@...>
 

Well, if you pull down the SXT frames and compare them to the Venice the exposure difference looks to be about a stop. That's not a good way to work if you're comparing cameras.

And if this is meant to be an "artistic" test then don't use charts that have known and standardized values, as people expect them to mean something. (Even Geoff's cool saturation chart has white, black and middle gray on it.) Just shoot a scene. 

--
Art Adams
DP
San Francisco Bay Area



On Mar 26, 2018 at 3:56 PM, <Michael Nagel via Cml.News> wrote:

50% IRE for mid-grey used to be the "standard reference".

there's nothing here that "works" because this is a non-scientific test. He's just leveling the playing field by doing the same thing for all cams involved. (at least this is how I understand it)

The other thing is, since you don't know everybody's viewing environment, where would you place mid-grey ? How would you know whether a 2.2 or 2.4 gamma assessment is "correct"... ? I've calibrated screens for peeps with a 2.6 gamma, worked just fine for them and their environment... you don't whether the user is viewing on CRT/LCD/LED/OLED/RPTV or projectors...

Also, the mid-grey chip gain adjustment he does is just to ballpark the actual EI/ISO of the cam. Again, non-scientific. This is just to see whether there is a strong deviation from what the manufacturer claims. At least that's how I understand his workflow.

the over/under frames are to see how color/texture holds up and when/how noise gets introduced. IMO, this is kinda drastic on some cams, see also the 2015 tests. Now, obviously you can address some of that in post, but that was not the point of the comparison.

all of the over/under have the same reference point (0 exposure), whether you think that's "too bright" or not doesn't really matter - this is all relative - since nobody knows the true EI of the cam and the manufacturer's claims can NEVER be trusted. ;-)

- Mike Nagel
Director/Producer
L.A.


Re: Venice evaluation

Mike Nagel
 

50% IRE for mid-grey used to be the "standard reference".

there's nothing here that "works" because this is a non-scientific test. He's just leveling the playing field by doing the same thing for all cams involved. (at least this is how I understand it)

The other thing is, since you don't know everybody's viewing environment, where would you place mid-grey ? How would you know whether a 2.2 or 2.4 gamma assessment is "correct"... ? I've calibrated screens for peeps with a 2.6 gamma, worked just fine for them and their environment... you don't whether the user is viewing on CRT/LCD/LED/OLED/RPTV or projectors...

Also, the mid-grey chip gain adjustment he does is just to ballpark the actual EI/ISO of the cam. Again, non-scientific. This is just to see whether there is a strong deviation from what the manufacturer claims. At least that's how I understand his workflow.

the over/under frames are to see how color/texture holds up and when/how noise gets introduced. IMO, this is kinda drastic on some cams, see also the 2015 tests. Now, obviously you can address some of that in post, but that was not the point of the comparison.

all of the over/under have the same reference point (0 exposure), whether you think that's "too bright" or not doesn't really matter - this is all relative - since nobody knows the true EI of the cam and the manufacturer's claims can NEVER be trusted. ;-)

- Mike Nagel
Director/Producer
L.A.


Re: Venice evaluation

Art Adams <art.cml.only@...>
 

Yeah, but there's always some sort of gamma applied, which means middle gray is not the same as splitting the code value range in half—not in Rec 709, or P3, or any log curve, or linear... not in anything.

You *can* put 18% gray wherever you want, but if you don't reproduce it in a way that resembles 18% gray on a display or screen of some sort, it's not 18% gray. The whole point is that it is a standard reference. If you don't use it that way, it loses its value. For example, when the ID on the EXR frames say "+5" or "-2"... in relation to what, if middle gray isn't middle gray? And isn't this deceptive if one just wants to look at the 0% frame and see how it grades? Doesn't it affect one's perception of noise in all the other frames?

And isn't judging how a camera looks all about perception?

If middle gray ever came to 50 IRE then those days are long gone by over a decade.

So sure, you *can* grade all this footage so middle gray is perceptually brighter than it should be, but then why include exposure information? It doesn't really relate to anything other than "plus or minus f/stops in relation to how someone thinks this footage looks good on Vimeo." That's not really a standard, and makes it a little more difficult to interpret the results across other display methods.

(I'm usually in the minority when it comes to this stuff, so if this works for you, cool. For me, it isn't a clean way to examine a camera.)

--
Art Adams
Director of Photography
San Francisco Bay Area


Re: New Red sensor.

Colin Elves
 

Oooh. Look. Red have a new sensor. And a new video featuring someone who doesn’t appear entirely comfortable with what they’ve been told to say.

image1.png

Slightly uncomfortable viewing for all!


Colin Elves
Director of Photography
London/Berlin


Re: Venice evaluation

Mike Nagel
 

On Mon, Mar 26, 2018 at 08:08 am, Art Adams wrote:
My color picker app shows middle gray in your Quicktimes to be 128, and you say you are grading middle gray to put green's code value at 512. Based on all of the above, I can't find any reason to believe that this is perceptually correct for display on a Rec 709 display, or an sRGB computer monitor. What was your rationale for choosing 512?
128 is 50% in 8-bit, with a max value of 256. 512 - which is what Geoff is targeting - is 50% of 10bit scale, with max value of 1024. Resolve shows waveform in 10bit, so if he adjusts in Resolve he's targeting 512, the 50% value.

Why 50% ?

here's what I assume: Geoff comes from film and back in the day an 18% Kodak grey card should equate in video to (roughly) 50% IRE. Now, u can still place mid grey where ever you want in your scene, but if he uses that as his target then that is why he is using these values.


Re: Venice evaluation

Art Adams <art.cml.only@...>
 

>The real problem is that no matter how well educated you become or how well calibrated you are, in my experience all that matters is how it looks on the random corporate issued laptop that the Creative Director reviews it on.   We literally had to regrade a piece once because a VP of marketing saw it on air on his home TV and decided the blacks were too dark!

Well... yes. I deal with that as well. But I prefer to start out knowing what I'm getting, as otherwise I feel like I'm flying blind.

I'm sure there are lots of people out there shooting with monitors in various configurations. I want to be the one with a monitor that is set up as properly as possible. No matter what it looks like on the client's Dell PC at home, at least I have a chance at it still looking good if I made it look that way in the first place, and to some sort of standardized reference.

A significant number of my projects aren't graded at all anymore, or they are "graded" by the editor who slaps on a default LUT. At least I know what it will look like in the end.

I've always been a believer in more education than less. In some cases that's a significant part of what separates me from my competition: my results will be a bit more predictable, and that's the most important thing.

--
Art Adams
Director of Photography
San Francisco Bay Area


Re: Venice evaluation

Philip Holland
 

This is where the juice gets harder to squeeze for sure.


One of the last HDR projects we final graded at Dolby on the Pulsar here in Burbank, but even still nailing a few color targets were interesting with the various parties involved.  I personally have shy of a dozen displays I QC and review on for 3 of my clients.  Some of those displays don't come close to achieving Rec.709, some cover it fully.  The harder target sometimes is providing a balanced solution to final color that works broadly rather than on perfectly calibrated displays.  Meanwhile at the same time consumer displays are becoming better and in some cases rather good.


Side note and off topic, though likely of interest to some here.  I was at the MITC meeting at the ASC last week and there were a broad range of topics covered.  Specifically to note are the efforts going into ACES Next which should right some of the wrongs on that front and improve things overall, hopefully workflow side too.  Some lively discussion regarding HDR10 and Dolby Vision as well.  I was keen to point out the newer ATSC 3.0 standards which were approved last November, which will manifest in consumer priced televisions this year with provisions for 4K, HDR, and 8K through extension.  Which means as of this year via streaming, broadcast, and satellite there's support at a minimum for 4K HDR in the home with one streaming 8K effort by the end of the year.  Theatrically a lot of focus has been put on 4K HDR/8K HDR via non-projection based systems as well (i.e. cinema LED).  This actually makes a great deal of sense to combat against some of the issues that are inherent to projection in regards to specifically HDR.


Phil


-----------------
Phil Holland - Cinematographer
http://www.phfx.com
http://www.imdb.com/name/nm0390802/
818 470 0623


From: cml-raw-log-hdr@... <cml-raw-log-hdr@...> on behalf of Gavin Greenwalt <im.thatoneguy@...>
Sent: Monday, March 26, 2018 12:42:28 PM
To: cml-raw-log-hdr@...
Subject: Re: [raw-log-hdr] Venice evaluation
 

"Well, I learn new stuff every day. Thanks, all, for the education about 2.4 and BT1886.” – Art

The real problem is that no matter how well educated you become or how well calibrated you are, in my experience all that matters is how it looks on the random corporate issued laptop that the Creative Director reviews it on.   We literally had to regrade a piece once because a VP of marketing saw it on air on his home TV and decided the blacks were too dark!

Call me cynical but in my opinion anything in the vaguely 2.1 – 2.4 range will be as correct or wrong on any random viewer’s set as source material on a perfectly calibrated display. Not to mention you ship the spot to a company which will then distribute the ad to the stations (sometimes on tape).  I’ve seen them change the gamma and black\white points.  And then each tv station decides how they want to broadcast it.   And then it reaches the consumer’s TV which is probably on MAX INTENSITY FOOTBALL mode.   Then you have to deliver it to Vimeo\Youtube who also mess with the gamma differently and that ignores what web browser or app someone uses to watch Youtube, some of which will apply a rec709 to sRGB conversion… sometimes.   If it’s not in a theater or on a Bluray you can almost guarantee it’ll be wrong by the time it makes it to broadcast (at least in the commercial world).

I’m most looking forward to the HDR era because we can deliver a spot with defined NIT values like Dolby Vision\HDR10 and then it’s up to the display manufacturer to ensure the display is delivering the content to the best of the display and ambient room’s ability.     For broadcast TV the analog guys still seem to rule the roost where a signal is a signal.  All of the digital ecosystems have no excuse and should define how it was mastered and then if your display is setup for 2.4 gamma and the image is flagged as 2.2 it should convert internally.  Even if the content is mastered for standard dynamic range and rec709 content I think it should all be delivered in HDR10 or Dolby Vision containers so that there is no ambiguity or interpretation on how it was intended to be viewed.   

That’s ultimately why you shouldn’t be taking any of the H264 clips seriously regardless if it was encoded in x264 or viewed on Vimeo or YouTube.   I can open two different browsers on my one monitor and get two completely different images on the same Vimeo page.   So we also would need to know what browser and OS Geoff is deciding that his vimeo output looks closest to his intent.   Then if the OS is applying any ICC color magic in the background we also need to know what monitor or device it looks most pleasing in.  What was the old back-ronym for NTSC “Never Twice the Same Color”? 50 years later and your average web video isn’t really any better.

Gavin Greenwalt
VFX

Seattle, WA

1701 - 1720 of 1984