Pages: 1 [2]   Go Down

Author Topic: How good is camera profiling really?  (Read 24644 times)

Fips

  • Full Member
  • ***
  • Offline Offline
  • Posts: 195
    • some unrelated photos on flickr
Re: How good is camera profiling really?
« Reply #20 on: January 12, 2013, 07:28:15 am »

Quote
He reckons it is inferior to the Adobe product but a must have? I see a contradiction that put any thoughts of spending £70 out of my mind. I don't really see the point of setting accurate colour and then possibly changing the hue, saturation or contrast of a particular colour in LR/ACR . If you are saving for web then it is a pointless exercise?

That's not how I read it. It might(!) be inferior for certain cameras and under certain conditions where lighting is well behaved. Like I said, my experience is that under difficult conditions I get more natural results using my custom profiles. Another example where Adobes profiles won't help you are polarizers. That yellowish tint I get from my Hoya HD under certain angles is almost impossible to get rid of using HSL alone.
I agree that if you do heavy post processing or even convert to B/W than there's no point in using custom profiles. But even for more moderate darkroom-like processing, I find it easier to get better looking results when I start from natural colors.
Logged

SZRitter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384
Re: How good is camera profiling really?
« Reply #21 on: January 17, 2013, 12:17:49 pm »

So, being lazy, I haven't ordered one yet. But now I am leaning towards the Classic and not the Passport.

I am guessing the larger size will make it easier to shoot, and it will fit perfectly in the back pocket of a LowePro shoulder bag I have. It's cheaper, which means more money towards some of the things I need (spare battery and hood/filter adapter for the X10, Lightroom or Capture1, softbox).

Any reason this isn't a good idea?
Logged

Fips

  • Full Member
  • ***
  • Offline Offline
  • Posts: 195
    • some unrelated photos on flickr
Re: How good is camera profiling really?
« Reply #22 on: January 17, 2013, 12:26:04 pm »

Quote
Any reason this isn't a good idea?

Depends. The advantages of the Passport over the Classic are:
  i) fits in a coat pocket,
  ii) when closed, the color patches are protected from a) UV and b) scratches,
  iii) large grey card on the back,
  iv) build in stand, so you don't need someone to hold it for you.
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: How good is camera profiling really?
« Reply #23 on: January 17, 2013, 01:03:58 pm »

If you go down the road of " accuracy" and produce a good print and then show it to somebody how do they the colours are "accurate" and further more do they care? I am excluding people who have commissioned a photographer to shoot some fashion clothing or similar.
I shoot images of the kids. My (non-photographer) significant other would periodically complain that the color of a dress or a rain-suit are "wrong", even though my Dell 2711 wide-gamut display was profiled and calibrated using the eye1 d2, and my 9000mk2 printer was used with original paper manufacturer profiles.

After profiling my 7D with a colorchecker, those complaints have become more seldom.

-h
Logged

SZRitter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384
Re: How good is camera profiling really?
« Reply #24 on: January 17, 2013, 01:29:09 pm »

Depends. The advantages of the Passport over the Classic are:
  i) fits in a coat pocket,
  ii) when closed, the color patches are protected from a) UV and b) scratches,
  iii) large grey card on the back,
  iv) build in stand, so you don't need someone to hold it for you.

i) True. I'm lazy enough with CC sized grey cards.... Not sure even if it's that small I'll carry it. That said, maybe I will, especially on paid shoots where I need better colors.

ii) Also a very valid point

iii) It's on the back? Huh, never noticed it in a photo, I figure they were just talking about the patches...

iv) Extremely good point. That right there could make it more useful. If I'm hiking or solo shooting an event, it can be tough/impossible to have someone hold a card for me.

Stupid choices.... Always making life complicated....
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: How good is camera profiling really?
« Reply #25 on: January 17, 2013, 02:18:06 pm »

If I had neither, I'd go for the Passport instead of the Classic in a second. The Classic is large but fragile. The Passport has those great off-white patches for setting WB. You get a larger white and gray card too. All protected in a case.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6229
    • http:www.schewephoto.com
Re: How good is camera profiling really?
« Reply #26 on: January 18, 2013, 12:08:58 am »

If I had neither, I'd go for the Passport instead of the Classic in a second. The Classic is large but fragile. The Passport has those great off-white patches for setting WB. You get a larger white and gray card too. All protected in a case.

+1

The problem with the regular ColorChecker (even the small one that is discontinued I think) is that it's very delicate and has a short shelf life (they will only last several years at best). The Passport has a hard outer case that protects the CC card as well as offered additional targets such as warming/cooling WB as well as color samples that are custom tuned to match up with the 8 hue colors in Camera Raw/Lightroom's HSL sliders.

Seriously, if you want to have a handy target for making profiles, you would be an idiot to buy the full large CC card and ignore the other bennies of the Passport.

The key thing you need to understand regarding the differences between using the Adobe Standard DNG profile and making your own custom DNG profile is that it's quite possible that YOUR sensor (the one in YOUR camera) is slightly different in it's spectral response from the camera/sensor Adobe used to create Adobe Standard. Therefore the custom DNG profile may be more accurate for YOUR camera...

Bruce Fraser and I tested two cameras (the original Digital Rebel) and found substantial differences between the two lots of cameras...we had different spectral responses and different native ISOs. His camera was about 1/3 stop hotter than mine and our cameras needed slightly different camera calibration settings.

Each sensor lot is bound to have individual and slightly different responses...if you value the best potential capture, you would do well to do a custom profile of your camera's sensor. The same capture can be used in the Passport software and the free DNG Profile Editor...if you are anal, you can try creating a profile in either–both will use the same targets (full size CC or Passport). Note that the DNG Profile Editor allows further editing of the resulting profile.

Also, be sure you understand the purpose of a DNG Profile...it's not a color editing tool to take the place of properly adjusting the image colors, it's a color rendering tool to allow you to start off with an optimal color rendering that you can then adjust further.

One is a technical tool and the other is the color correction tool...DNG Profile for technical, the ACR/LR sliders for color correction.
« Last Edit: January 18, 2013, 12:11:02 am by Schewe »
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: How good is camera profiling really?
« Reply #27 on: January 18, 2013, 02:09:56 am »

I was not aware that two different rebels could have significantly different spectral response. Perhaps Canon secretly update its sensor manufacture process (or only cfa) during a product lifecycle (to save cost, or to replace a bad supplier). In that case, one would expect there to be two or "a few" different responses out there, within which the response should be very similar?

The counter argument would be that Adobe & friends surely have more extensive testing labs than most of us, and surely employ better color scientists than myself. They could base their camera profiles on large sets of patches illuminated by controlled sources that are deemed relevant for most scenes. The best I can (will) do is my camera, the passport checker, sunlight/shady/direct flash/flash in my living-room celing/tungsten.

Yet my experience is that (especially) saturated man-made red colors are reproduced with a more convincing hue using my profiles than any of the canned profiles.

I believe that if the camera spectral response deviate substantially from the ideal (to save cost, to improve luminance SNR or something else), the color correction may be more nonlinear, and sensitive to measurement errors, irregular illumination etc.

-h
Logged

SZRitter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384
Re: How good is camera profiling really?
« Reply #28 on: January 18, 2013, 11:50:30 am »

Thanks everyone. I'll be ordering the Passport on Saturday.
Logged

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6229
    • http:www.schewephoto.com
Re: How good is camera profiling really?
« Reply #29 on: January 19, 2013, 01:33:27 am »

The counter argument would be that Adobe & friends surely have more extensive testing labs than most of us, and surely employ better color scientists than myself. They could base their camera profiles on large sets of patches illuminated by controlled sources that are deemed relevant for most scenes. The best I can (will) do is my camera, the passport checker, sunlight/shady/direct flash/flash in my living-room celing/tungsten.

Yeah, ya know...it's one guy based in Boston (for the most part) who ends up getting preproduction or early production model cameras and does a series of tests on a small sample of cameras (generally one) and while to tools that guy uses are pretty exotic and accurate, it's only for a single (generally) unit camera. How much a camera will vary lot to lot is a question for the camera makers (good luck getting any data) but if you know anything about statistics and process control you would know that there will be differences from lot to lot of sensors–that's almost assured of happening.

Since the specific camera support always happens early in a camera's life, you can be pretty assured that second-gen releases will vary a bit from the initial round of manufacturer releases...if the camera is really successful, there may be many multiple of rounds or releases. It would be really foolish to expect multiple lots of sensors to preform exactly the same.

Over the years, ACR/LR have had multiple rounds of camera calibrations made available. Any camera you have that shows more than one profile has had an "adjustment. Since the advent of the DNG profiles, Adobe (meaning our friend in Boston) may take another look at a camera profile after the initial camera support was added. In general, when that happens, Adobe doesn't make an announcement but simply modifies the Adobe Standard profile that ships with a particular version of ACR/LR. Note, that doesn't happen a lot.

But it's not at all unusual nor unexpected that a camera and it's sensor may be slightly (or more) different than the unit that Adobe tested and made the initial DNG profiles for. For that reason, it's useful to shoot your own CC targets and use either the Passport software or the DNG Profile Editor to create DNG profiles. Then you can compare and contrast the Adobe Standard to the custom profile you made from YOUR sensor and see which one handles color the best.

I've had cameras where there was no appreciable difference and I've had cameras where the custom profile was considerably "better". You really can't know until you do your own testing after making your own DNG profile.
Logged

mac_paolo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 431
Re: How good is camera profiling really?
« Reply #30 on: January 19, 2013, 05:51:30 am »

I agree with Jeff.

Not only our Boston based guru often tests one camera but, as Jeff mentioned, sometimes it's a preproduction model and in the years, the way Adobe Standard Profiles are made has changed, thus for old cameras (>3 years), custom built profiles may differ a lot.
My old and loved D300 is waaaay better with the custom profile over standard one.
My little and new RX100 is somewhere better, more accurate for sure, yet not night and day as for the former.

Some people tend to tell that's overkill to custom profile a camera. If you are serious enough into photography I -strongly- recommend to buy a Passport/CC24 and make at least a basic dual illuminant profile.

To tell the truth I'd prefer to be able to use a CC Digital SG, but that's another story :)
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: How good is camera profiling really?
« Reply #31 on: January 19, 2013, 11:29:49 am »

The key thing you need to understand regarding the differences between using the Adobe Standard DNG profile and making your own custom DNG profile is that it's quite possible that YOUR sensor (the one in YOUR camera) is slightly different in it's spectral response from the camera/sensor Adobe used to create Adobe Standard. Therefore the custom DNG profile may be more accurate for YOUR camera...

and you forget that the difference in your sensor/CFA/lenses are negated by the difference in your copy of the target vs averaged patch data hardcoded in software and by the difference in quality of spectral readings using calibrated monochromator vs poorly illuminated 24 patch target (and some dudes who shoot target filling the whole frame w/ it)...  the mere fact that Adobe itself does not use neither the target nor software (DNG PE) simply tells that it is mostly a snake oil for a regular illumination... if you do not like the rendering then get Adobe DNG PE and just tune the profile to your taste (or use Sandy Mc tools and simply get rid of LUTs and insert a linear tone curve, matrix only is good enough) w/o any target shooting.


Bruce Fraser and I tested two cameras (the original Digital Rebel) and found substantial differences between the two lots of cameras...we had different spectral responses and different native ISOs. His camera was about 1/3 stop hotter than mine and our cameras needed slightly different camera calibration settings.

and that was when consumer level (original Rebel) dSLRs were in their infancy... why didn't you repeat the same w/ modern cameras ?

now if you can get an individually calibrated spectrophotometer (and not just the one purchased and never actually calibrated to check that it measures properly) and measure two targets what do you find  ;) ?

but placebo effect exists
Logged

mac_paolo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 431
Re: How good is camera profiling really?
« Reply #32 on: February 28, 2013, 01:28:06 pm »

and that was when consumer level (original Rebel) dSLRs were in their infancy... why didn't you repeat the same w/ modern cameras ?

now if you can get an individually calibrated spectrophotometer (and not just the one purchased and never actually calibrated to check that it measures properly) and measure two targets what do you find  ;) ?

but placebo effect exists
Can't grasp your point.
Are you telling that modern DSLR sensors are all almost equal and thereafter the well built Adobe Standard profile is the only good choice over custom ones?
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: How good is camera profiling really?
« Reply #33 on: February 28, 2013, 01:32:28 pm »

Are you telling that modern DSLR sensors are all almost equal and thereafter the well built Adobe Standard profile is the only good choice over custom ones?

if you want to talk about your personal tastes then please by all means enjoy profile building... otherwise make a pure matrix profile (do not forget to remove LUTs from yours and Adobe Standard) and show us dEs vs a different target under a regular light (no fluorescent, sodium vapors, etc).
Logged

l_d_allan

  • Full Member
  • ***
  • Offline Offline
  • Posts: 208
    • Berean photos
Re: How good is camera profiling really?
« Reply #34 on: March 04, 2013, 04:20:58 pm »

Yeah, ya know...it's one guy based in Boston (for the most part) who ends up getting preproduction or early production model cameras and does a series of tests on a small sample of cameras (generally one) and while to tools that guy uses are pretty exotic and accurate, it's only for a single (generally) unit camera. How much a camera will vary lot to lot is a question for the camera makers (good luck getting any data) but if you know anything about statistics and process control you would know that there will be differences from lot to lot of sensors–that's almost assured of happening.

The name Roger Cicala (founder of LensRentals) comes to mind. He has Lots of experience measuring a decent sample size for performance. He really seems to enjoy that, and be good at it.

LensRentals probably has dozens if not hundreds of a specific model, like a bunch of Canon 5dm3's or Nikon D800's. He's in a position to measure a more statistically valid sample size (1000% to 5000% greater than one sample).

I'd find it interesting to see standard deviations from the Adobe Profile, as well as within the same camera model. This could be drilled down to the 3x8 individual slider values for HSL Reds, Oranges, ... Purples, and Magenta.
« Last Edit: March 04, 2013, 04:33:07 pm by l_d_allan »
Logged
retired in Colorado Springs, CO, USA ... hobby'ist with mostly Canon gear

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: How good is camera profiling really?
« Reply #35 on: March 05, 2013, 04:43:05 am »

Is there a good way to present (a condensated) difference between two profiles?

I have profiled my ND/circpol filters, and my lenses on the same body, using the same colorchecker passport and under the (nominally) same lighting. I can use those profiles in Lightroom, and I see some difference, not something that is very important to me.

It would be nice to be able to say that "yeah, my sample of the Hoya 400x ND has a WB shift of +200K, or stated differently, a HSL-correction of xyz makes it neutral under shady daylight".

It would also be interesting to see quantitatively what separates my 7D profiles from Adobe and Canon ones. Perhaps visualizing the parametric "point-cloud" can tell us something about what is going on except measurement noise?

-h
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: How good is camera profiling really?
« Reply #36 on: March 05, 2013, 10:11:31 am »

Is there a good way to present (a condensated) difference between two profiles?

What I'd do** is capture a target (with as many patches as I could find) with both profiles. Then in Photoshop, I'd do my best to remove anything that isn't patch data by cropping. Then I'd resample down using nearest Neighbor and feed a small file to ColorThink Pro, asking it to extract all unique colors from both. That would provide the source to save out a Color List which could then be compared to the other color list in CTP and provide a dE report (Average, Max/min, std Dev etc).

Or you could just toggle between the two and pick the one you prefer <g>. But if you want numbers, CTP could provide them.

** http://www.luminous-landscape.com/forum/index.php?topic=75480.0
« Last Edit: March 05, 2013, 06:08:10 pm by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

l_d_allan

  • Full Member
  • ***
  • Offline Offline
  • Posts: 208
    • Berean photos
Re: How good is camera profiling really?
« Reply #37 on: March 05, 2013, 10:37:05 am »

Is there a good way to present (a condensated) difference between two profiles?

If you've got i1Profiler or ProfileMaker, they each have a "Compare" capability. I've only used PM5 + Compare with print profiles, but my understanding is that you can also compare camera profiles. Patch by patch, and overall.

PMP5 + Compare works in demo mode even without a dongle. IIRC, that's not the case with i1Profiler.
Logged
retired in Colorado Springs, CO, USA ... hobby'ist with mostly Canon gear

Ellis Vener

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2151
    • http://www.ellisvener.com
Re: How good is camera profiling really?
« Reply #38 on: March 07, 2013, 01:43:22 pm »

To get out of the technical weeds and to actually respond to Fips question, which was:

"How well does profiling your camera really work?"

I have made profiles for several different cameras (about a dozen) using different tools, most recently and in alphabetical order, Adobe's DNG profile editor (with a 24 patch Xrite ColorChecker); Datacolor's SpyderCheckr; and Xrite's ColorChecker Passport. 

Some cameras (and brands)  "need" it more than others to get them closer to an overall "neutral" rendering, but having done it multiple times in multiple ways on multiple cameras, my take on camera profiling is that it serves too purposes:

-Trying to create an as close to accurate as you can get color rendering for copywork.

or

- Creating an additional palette for  creative work. Sometimes I find it very useful to  have different  starting points for how I want to interpret  the colors in a photograph   to create the emotional and aesthetic effect in the viewer's mind to help shape what they think about the subject and the photograph. 

If you are old enough to remember film,  the analogy is to how I once would choose a film stock to work with on a project: Velvia vs. Kodachrome vs. Provia vs.  E100S vs. E100Vs vs EPR vs Portra vs. VPS vs. Reala vs. etc.
Logged

NancyP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2513
Re: How good is camera profiling really?
« Reply #39 on: March 25, 2013, 06:26:19 pm »

I guess I will find out soon. After looking at another photographer's Adobe standard DNG vs custom CC profile output and seeing interesting color separation with the latter, I have decided to experiment. CC is coming this week, and I can get out and shoot under various conditions.
Logged
Pages: 1 [2]   Go Up