jotdeh wrote:
I just followed the instructions in this thread.
Take a fully saturated (overexposed image), say 30 sec, and then use this as the basis for a custom white balance. My 5DII tells me that this image "may not be suitable for custom WB" but actually what happens if you do is that the camera's built-in channel response correction is reversed, and your jpgs will show a much better representation of what the sensor really captured in each channel. You will see a strong green cast - basically the reds are severely underexposed compared to greens if you shot a gray card. The camera pushes the reds by something like 1.5 stops!! Now if you used a red filter (or magenta as noted, because blues are weaker than greens, too) then you can really balance the response of each channel. ...Show more →
Oh, UniWB and linear gamma and all that :-) I've played with it a little bit, but it essentially makes in-camera JPGs very weird, so looking at your camera's LCD stops being useful. That's a heavy price to pay for advantages that are more theoretical than practical. I decided it's not worth it for me.
If you want to see what the sensor actually captures in each channel, there's a variety of software (RawTherapee, etc.) which will show it to you. And yes, the green sensels get considerably more photons that red sensels, but so what? It's not like you'll replace a red subject with a green subject to reduce noise :-D
KaaX wrote:
Oh, UniWB and linear gamma and all that :-) I've played with it a little bit, but it essentially makes in-camera JPGs very weird, so looking at your camera's LCD stops being useful. That's a heavy price to pay for advantages that are more theoretical than practical. I decided it's not worth it for me.
If you want to see what the sensor actually captures in each channel, there's a variety of software (RawTherapee, etc.) which will show it to you. And yes, the green sensels get considerably more photons that red sensels, but so what? It's not like you'll replace a red subject with a green subject to reduce noise :-D
...Show more →
Well, if you use a magenta filter than the jpg should look pretty natural again. But I agree, you can get beautiful images without worrying about somewhat reduced dynamic range in the red and blue channel.
jotdeh wrote:
Well, if you use a magenta filter than the jpg should look pretty natural again.
Yeah, but that's just pushing things towards the lowest common denominator -- your red and blue sensels are worse than green ones so by using the magenta filter you intentionally downgrade your green response to the bad level of red and blue. You're throwing out good information for no good reason.
And again, we know that we can re-balance the R, G, and B channels to produce a adequate color rendering. So outside of colorimetric applications I'm not sure that UniWB is all that useful. You discard light for minor theoretical advantages.
KaaX wrote:
Hmm... interesting -- but why do you do this? Purely to make the image you see on the LCD on the back of the camera more "standardized"? I pretty much always adjust the WB by hand in post-production so I don't really care about the WB during shooting and have it set on Auto almost all the time. Sometimes it misses pretty badly, but I would expect AutoWB to be a better "foundation" than, say, "cloudy" -- all your sunlit images will start out too warm, for example... Is there something I'm missing?
It's about developing a "look" to your shooting. Like with film, using a static WB most of the time gives me a look to start with, and I can always adjust after the fact. I think digital shooters can often be too concerned about correct WB vs. nice looking WB, and we end up with a bunch of homogenous shots that look the same despite being taken in very different lighting scenarios at different times of day. Plus, autoWB can often give pretty strange starting points in contrasty outdoor scenes.
So, I think of it as using a single film, but, rather than needing to use filters on the lens, I can correct the color after the fact if need be, but I often don't need to do much. Granted, despite being calibrated to a single color temp, film seems to have a better range in going from a cooler to warmer outdoor scene. That's why I tend to use closer to 6K, rather than 5K, with my camera, because it think it does bridge that gap a little better. I do occasionally "change my film" though and switch to a cooler starting point. I know that one can select every WB possible for every photo, until the "right" strikes his fancy, but it can be hard to target a direction without a consistent starting point.
Of course, if you're shooting products professionally, or something similar, then it's a very different thing, but I'm talking about in terms of personal or fine art shooting. Shooting with a consistent WB gives your digital images a slightly more "film-like" vibe, if I dare say such a thing.
KaaX wrote:
Yeah, but that's just pushing things towards the lowest common denominator -- your red and blue sensels are worse than green ones so by using the magenta filter you intentionally downgrade your green response to the bad level of red and blue. You're throwing out good information for no good reason.
And again, we know that we can re-balance the R, G, and B channels to produce a adequate color rendering. So outside of colorimetric applications I'm not sure that UniWB is all that useful. You discard light for minor theoretical advantages.
You're not dragging the green channel down to the same "bad" as the red - you allow the red more time to compensate for its weakness. What you lose in quality is that your shutter time slow down 1.5 stops. That may or may not matter, and it may or may not give a significant enough colour quality improvement. It depends. But I'm intrigued to try this, especially for flower shots.
KaaX wrote:
Yeah, but that's just pushing things towards the lowest common denominator -- your red and blue sensels are worse than green ones so by using the magenta filter you intentionally downgrade your green response to the bad level of red and blue. You're throwing out good information for no good reason.
And again, we know that we can re-balance the R, G, and B channels to produce a adequate color rendering. So outside of colorimetric applications I'm not sure that UniWB is all that useful. You discard light for minor theoretical advantages.
The magenta filter allows you to bring up the exposure in the R and B channels by attenuating the G channels, which will potentially result in less shadow noise and an expansion of usable DR.
douglasf13 wrote:
It's about developing a "look" to your shooting. ... I think digital shooters can often be too concerned about correct WB vs. nice looking WB, and we end up with a bunch of homogenous shots that look the same despite being taken in very different lighting scenarios at different times of day.
Ah. That makes some sense. I hear you about forcing each shot to the "correct" WB, that's common trap to fall into.
I still think that the goal is to get in sync with the color response of the human vision (which is highly adaptable but within limits), not with the limitations of film, but hey, to each his own :-)
jotdeh wrote:
You're not dragging the green channel down to the same "bad" as the red - you allow the red more time to compensate for its weakness. What you lose in quality is that your shutter time slow down 1.5 stops. That may or may not matter, and it may or may not give a significant enough colour quality improvement. It depends. But I'm intrigued to try this, especially for flower shots.
douglasf13 wrote:
The magenta filter allows you to bring up the exposure in the R and B channels by attenuating the G channels, which will potentially result in less shadow noise and an expansion of usable DR.
I see. So compared to the "normal" case you overexpose and use the magenta filter to keep the green sensels from blowing out. Hmm...
I'm still not sure I'm buying the "less shadow noise" argument. If you're shooting at base ISO, the shadow noise (at least on contemporary Sony/Nikon sensors) is not an issue at all. And if you're not shooting at base ISO, then you pay for balanced R-G-B exposure by having to increase your ISO by 1.5 stops which, I think, would kill all and any shadow noise advantage.
Basically, if you have more than enough light, you're solving a problem which isn't a problem. And if you don't have enough light, you're paying by increased ISO which makes the exercise pointless.
KaaX wrote:
I see. So compared to the "normal" case you overexpose and use the magenta filter to keep the green sensels from blowing out. Hmm...
I'm still not sure I'm buying the "less shadow noise" argument. If you're shooting at base ISO, the shadow noise (at least on contemporary Sony/Nikon sensors) is not an issue at all. And if you're not shooting at base ISO, then you pay for balanced R-G-B exposure by having to increase your ISO by 1.5 stops which, I think, would kill all and any shadow noise advantage.
Basically, if you have more than enough light, you're solving a problem which isn't a problem. And if you don't have enough light, you're paying by increased ISO which makes the exercise pointless.
I've really only heard of magenta filters and uniWB used by landscape shooters with tripods at base ISO. I agree that the newer sensors have made this method less useful, but you'll still get cleaner shadows, which could make a difference to some who benefit from wringing out the most DR that they can. I've personally never had a use for it, and I don't think the technique would make sense for everyday, non-tripod shooting.
I venture to say the benefit is of the same order of magnitude as the "expose to the right". Because you have to push the reds you're not using the full response resolution your files could carry. If you use the magenta filter to equalise the channel response then you can do "expose to the right" for the red channel, too.
ettr is not necessary to get great photos. It is merely a technique to get the highest technical quality out of the system you have.
Similarly, using a magenta filter is not necessary, but a means for getting optimal performance in certain situations.
I love that I'm talking about this without having used the technique - very internetesque, me thinks
KaaX wrote:
I'm still not sure I'm buying the "less shadow noise" argument. If you're shooting at base ISO, the shadow noise (at least on contemporary Sony/Nikon sensors) is not an issue at all. And if you're not shooting at base ISO, then you pay for balanced R-G-B exposure by having to increase your ISO by 1.5 stops which, I think, would kill all and any shadow noise advantage.
Basically, if you have more than enough light, you're solving a problem which isn't a problem. And if you don't have enough light, you're paying by increased ISO which makes the exercise pointless.
What if you use a longer exposure time... I think it could help reduce things like grain in blue skies, esp. with landscape shots taken from a tripod (making compositions with a magenta viewfinder image won't be so nice anyway, so you could focus and make your composition before screwing on the filter).
@jotdeh: Some people say that ETTR isn't optimal in every way because a longer exposure for the same scene increases the contribution of lens flare, which lowers contrast and worsens tonality.
douglasf13 wrote:
I don't think the technique would make sensor for everyday, non-tripod shooting.
AhamB wrote:
@KaaX: Douglas shoots with a NEX-7 (no OVF), so the colors affect what he sees while making compositions.
I'm guessing he uses the Cloudy preset because he probably applies a warm film toning to most of his images in PP.
Not really. I don't shoot particularly warm. My setting is 6K, which is still defined by some as regular daylight. It's just that the "cloudy" setting in my particular camera, at 6K, is probably on the cool side of cloudy...or on the warm side of daylight. As I mentioned, I do sometimes use a cooler WB.
The WB "method" (if you can call it that) I'm using can be used with any baseline WB you wish.
Yeah, it's kinda HDR for sensor color channels :-) But anyway, it seems we've made a circle coming back to the original opinion: UniWB has theoretical advantages but it's too much hassle for too little gain in the real world.
But to get back (somewhat) on track, did anyone actually play with using camera profiles and/or VSCO plug-ins and/or DXO "film look" profiles, etc. to get some approximation to a specific film look on digital? How well does it work?
douglasf13 wrote:
A lot of shooters use UniWB and a magenta filter, which evens out the color response on digital and improves dynamic range.
Also, I think autoWB is a big culprit for digital shooters who want more consistent tonality. I learned from Jono Slack on another forum that keeping my camera set to a single WB allows me to visualize things a little easier and more consistently. When I'm outside, I always have my camera set to "cloudy" outdoors and "tungsten" indoors, no matter what the conditions, and that gives me a good foundation. I can always adjust the raw from there, if need be. I do still occasionally use autoWB in mixed lighting, but I try to keep my camera set at "cloudy" outdoors and "tungsten" indoors most of the time....Show more →
This is interesting. I recently came across this philosophy I think on Thorsten Overgaard's site, where he follows the same practice when going between indoor and outdoor work. He also automatically switches between ISO 800 indoors and 160 outdoors.
I think from what you explained I can appreciate your reasons for doing this. Film was frequently used in 'unmatched' environments where the existing colour temperature was not a perfect or good fit, and the results were characteristic and accepted as such (except by those with colour temperature meters and stacks of Wratten filters). When I studied photography, part of the technical training was to evaluate film stock characteristics and determine the appropriate filtration correction. IIRC, Kodak even published suggested filter corrections on a batch by batch basis for certain emulsions. Instructors regaled us with tales of buying cases of film and freezing it to lock in a consistent look for use over a span of time.
With digital, as discussed, there is a feeling of needing to set the perfect WB. I know going through my years of digital images I can see how my PP style changed, including my WB biases. I agree AWB can be misleading, but I think it comes down to each individual's PP experience and ability to assess colour. One of the WB 'errors' that irks me in some of the image threads is AWB in forest scenes, where the camera tries to overcompensate for the predominance of green by adding loads of magenta and dropping the K value, resulting in purple tree trunks. I always hated seeing this in prints from mini-labs, and it carries over into digital. I see it in my own images because I shoot everything in AWB and quickly reset it in post, likely because I was never forced to be a Jpeg shooter and therefore shoot like someone working with slide film.
Another interesting tidbit I think I read at Overgaard's site was the statement that Leica developed the look of the digital M's output to mimic that of Kodachrome. Interesting of course that the sensor also originated from Kodak. IMO Kodachrome is not an accurate colour palette, but certainly has a distinctive look. Just as I think Edward's opening image is a pretty good example of Fujichrome - a bit cool and magenta biased. I've read some opinions that the colour philosophy of Kodak vs. Fuji films was due to the idealized skin colour preferences of each company's primary markets - Kodak's being predominantly caucasian and Fuji's being asian.
But I think maybe we can agree that UniWB and magenta filters "have practical advantages for certain purposes" (i.e. landscape with non-grainy/noisy blue skies, as noted by AhamB) and that, absolutely right, it is not a cure for other types of shots, especially those where you need to get in light as fast as possible (i.e. mostly hand-held / action-stopping scenes). Also keeping in mind that this process may adds processing overhead that is not warranted by the increased tonal quality, or that the filter may be detrimental to sharpness, just like a UV filter.
AhamB, that increased flare / reduced contrast argument with ETTR has probably been exhausted elsewhere ("Some people say...", probably on the internet), but at first sight I would say that if your lens produces flare it will be there no matter what. If you ETTR and in post you bring the exposure back down for proper visual rendition you should be getting all the ETTR benefits with no more flare than you would've gotten with ordinary "proper" exposure.
But that's just my guess, others probably investigated properly.
jotdeh wrote:
AhamB, that increased flare / reduced contrast argument with ETTR has probably been exhausted elsewhere ("Some people say...", probably on the internet), but at first sight I would say that if your lens produces flare it will be there no matter what. If you ETTR and in post you bring the exposure back down for proper visual rendition you should be getting all the ETTR benefits with no more flare than you would've gotten with ordinary "proper" exposure.
But that's just my guess, others probably investigated properly.
@jotdeh: I'm talking about localized flare but global veiling flare due to internal reflections that are there simply due to the scene illuminance. The point was brought up in a discussion about ETTR in this forum. I don't remember who it was (might have been Douglas?), but someone said in that thread that ETTR produced inferior tonality with the Sony A900, which was attributed to veiling flare by the person.
Cool, I'd love to learn more about that - anything that is counterintuitive (to me) will be a great learning opportunity (for me) (corporate speak, haha!)
The makers of RPP used to talk about how exposing mid tones on the A900 (and probably any recent Exmor camera) too far to the right could negatively affect color. However, I believe that Joakim gave a counterargument to this, but I can't remember the details. I believe much of it has to do with the raw converter that you're using.
As I mentioned, I do sometimes use a cooler WB.