p.3 #2 · 5d iii help: horrible over saturated reds.
Ben,
While you're at it, could you please state what software and version you are using to develop the raw? LR4, v4.1, for example? Also, what camera profile are you using?
p.3 #7 · 5d iii help: horrible over saturated reds.
It's hard for me to come to a firm conclusion without seeing the raw.
What I find interesting is the shadowed part of the bumper, has roughly equal amounts of green and blue hence the red appearance. The lit portion, however, has a huge increase in the green channel, hence the yellow color. Now the question is why.
I am wondering if it might be the type of paint (fluorescent properties, for example), the cycling/type of the lights, or perhaps a combination of the two.
I also don't have 100% confidence in the DNG/LR3 combo for the 5D3. Correct me if I'm wrong but I think the camera profiles in LR3 are limited in this scenario. What profile are you using?
Sorry for so many questions but is this an isolated case or are you seeing this in other images with reds?
Could you give us the basic LR settings you used for the image as posted including temp and tint?
p.3 #8 · 5d iii help: horrible over saturated reds.
Eyeball wrote:
It's hard for me to come to a firm conclusion without seeing the raw.
What I find interesting is the shadowed part of the bumper, has roughly equal amounts of green and blue hence the red appearance. The lit portion, however, has a huge increase in the green channel, hence the yellow color. Now the question is why.
I am wondering if it might be the type of paint (fluorescent properties, for example), the cycling/type of the lights, or perhaps a combination of the two.
I also don't have 100% confidence in the DNG/LR3 combo for the 5D3. Correct me if I'm wrong but I think the camera profiles in LR3 are limited in this scenario. What profile are you using?
Sorry for so many questions but is this an isolated case or are you seeing this in other images with reds?
Could you give us the basic LR settings you used for the image as posted including temp and tint?...Show more →
the lit portion has a ton more green because it has a ton more light
the shadows maintain red because red is not blown so it maintains strong dominance over red and blue channels
in the lit part green goes way up, blue goes up, red can't go way up since it was already near max so you get like (using made up #s) 240,80,27 for shadows and then 255,245,80 for lit part (although in reality it like maybe 750,245,80 on the camera's scale as it was exposed but it can't capture and store that)
p.3 #9 · 5d iii help: horrible over saturated reds.
skibum5 wrote:
the shadows maintain red because red is not blown so it maintains strong dominance over red and blue channels
The red will always be some shade of red if the blue and green channels are roughly similar and less than the red.
skibum5 wrote:
in the lit part green goes way up, blue goes up
But the blue ISN'T going up. That is my point. The shadow areas have blue and green roughly equal. In fact, some portions of the shadows have blue slightly stronger than green.
Normally, a clip in the red channel only will still be red. You will just start seeing some loss of detail since the red channel is no longer contributing to the tonal variation. In this case, there is a shift to yellow and since neither the blue nor the green channel appear to be clipped, I don't see an obvious reason for the color shift.
Besides the possibilities that I already mentioned (paint/lighting), I wonder if there could be some cross-flooding of the camera's sensor elements. Another possibility could be the camera profile used but Ben says this is what he saw on the camera LCD so that tells me that the camera profile is not a likely culprit.
It could also indeed be just a result of significant clipping but I can't really remember seeing a strong shift to yellow like this, particularly when neither of the other channels appears to be clipped.
p.3 #10 · 5d iii help: horrible over saturated reds.
skibum5 wrote:
the lit portion has a ton more green because it has a ton more light
the shadows maintain red because red is not blown so it maintains strong dominance over red and blue channels
in the lit part green goes way up, blue goes up, red can't go way up since it was already near max so you get like (using made up #s) 240,80,27 for shadows and then 255,245,80 for lit part (although in reality it like maybe 750,245,80 on the camera's scale as it was exposed but it can't capture and store that)
Exactly (and basically what I tried to say without going into numbers). Although the OP has not noticed an effect like this with previous cameras, I "reminded myself" of having experienced a virtually identical effect with my 5D2 when I shot a fire engine at night. The scene was wide in DR, and although my exposures captured good shadow and highlight detail with appropriate histogram for a night scene, one surface of the red engine was facing me and strongly lit. I saw the 'blinkies" there but figured I had enough headroom in the raw files to recover. Nope. In fact, the red channel clipped so strongly that that area became yellow just like the railroad car in this thread.
I do think the evaluative metering of my 1D3 would have done a better job of closing in on an exposure that worked, but spot metering and/or more careful attention to the histogram would have taken care of it with the 5D2. I'm not seeing any problem with reds on my 5D3, but if I do, the first thing I'll try is tweaking my metering technique or settings. I hope it works out for the OP.
p.3 #11 · 5d iii help: horrible over saturated reds.
Eyeball wrote:
But the blue ISN'T going up. That is my point. The shadow areas have blue and green roughly equal. In fact, some portions of the shadows have blue slightly stronger than green.
Normally, a clip in the red channel only will still be red. You will just start seeing some loss of detail since the red channel is no longer contributing to the tonal variation. In this case, there is a shift to yellow and since neither the blue nor the green channel appear to be clipped, I don't see an obvious reason for the color shift.
Besides the possibilities that I already mentioned (paint/lighting), I wonder if there could be some cross-flooding of the camera's sensor elements. Another possibility could be the camera profile used but Ben says this is what he saw on the camera LCD so that tells me that the camera profile is not a likely culprit.
It could also indeed be just a result of significant clipping but I can't really remember seeing a strong shift to yellow like this, particularly when neither of the other channels appears to be clipped....Show more →
Hmm...if there is equal blue and green in that area, I'm stumped. Maybe it is partly a result of a light source reflecting strongly off that particular surface, despite the color of the areas seeming even. Or bleeding on the sensor. In any case, it does seem that red clipping is part of the issue, doesn't it?
p.3 #12 · 5d iii help: horrible over saturated reds.
Eyeball wrote:
The red will always be some shade of red if the blue and green channels are roughly similar and less than the red.
But the blue ISN'T going up. That is my point. The shadow areas have blue and green roughly equal. In fact, some portions of the shadows have blue slightly stronger than green.
Normally, a clip in the red channel only will still be red. You will just start seeing some loss of detail since the red channel is no longer contributing to the tonal variation. In this case, there is a shift to yellow and since neither the blue nor the green channel appear to be clipped, I don't see an obvious reason for the color shift.
Besides the possibilities that I already mentioned (paint/lighting), I wonder if there could be some cross-flooding of the camera's sensor elements. Another possibility could be the camera profile used but Ben says this is what he saw on the camera LCD so that tells me that the camera profile is not a likely culprit.
It could also indeed be just a result of significant clipping but I can't really remember seeing a strong shift to yellow like this, particularly when neither of the other channels appears to be clipped....Show more →
maybe blue doesn't go up since the color temp in the shadowed area is cool and the one in the lit area is warm and it happened to somehow balance it out more light but warmer vs less overall light but much cooler? That seems a bit surprising though I'd think direct sunlight would still put more blue onto it but maybe not.
p.3 #13 · 5d iii help: horrible over saturated reds.
Eyeball wrote:
The red will always be some shade of red if the blue and green channels are roughly similar and less than the red.
That is not the case. Imagine a really bad case in which there was some amount of green and blue in the subject, but that it was so badly overexposed that all three channels saturated - the result would no longer be blue at all. At some point short of that, it would be something barely red or pink... if the balance between the other two channels was equal. But in the case where it is not equal (in G and B) you will get something else.
Color shifts are, indeed, a result of blowing a channel.
p.3 #14 · 5d iii help: horrible over saturated reds.
gdanmitchell wrote:
That is not the case. Imagine a really bad case in which there was some amount of green and blue in the subject, but that it was so badly overexposed that all three channels saturated - the result would no longer be blue at all.
Agreed. If all three channels were blown, you would get white.
gdanmitchell wrote:
At some point short of that, it would be something barely red or pink... if the balance between the other two channels was equal. But in the case where it is not equal (in G and B) you will get something else.
Agreed, which is apparently what is happening here. My doubt is WHY the green channel is shooting up (but not clipping), particularly when the shadow area shows the blue and green channels roughly balanced. Seems to imply that the light sources are different. Ben is apparently having difficulty remembering the exact lighting scenario but it does not appear to be your typical outside orange sun/blue sky type of situation. Who knows?
gdanmitchell wrote:
Color shifts are, indeed, a result of blowing a channel.
I agree that a color shift may be a result of clipped channel, particularly if another channel is also clipped or the clipping is severe enough to change the RGB ratios significantly. The amount of the color shift in this case is what I find puzzling in this case, particularly with the huge boost in the green channel compared to the shadow areas.
p.3 #15 · 5d iii help: horrible over saturated reds.
Guys Dan is right, this type of color shift happens, and can be easily duplicated. Just try it yourself with something the proper shade of red if you do not believe.
5 different exposure of the same pamphlet in the same lighting, check the exif data and everything.
Notice how the dragon symbol changes from red to somewhat red-orange and finally yellow in the final shot as the over-exposure increases.
This was taken with a 5d2 by the way, not a 5d3. But it is by no means unique to Canon or DSLRs in general.
p.3 #16 · 5d iii help: horrible over saturated reds.
Access wrote:
Guys Dan is right, this type of color shift happens, and can be easily duplicated. Just try it yourself with something the proper shade of red if you do not believe.
Yes, it is possible for it to occur but you need two things for it to happen:
- A very orangish-red to begin with as you mention in your post.
- Really significant over-exposure of that red/orange color. Probably 3 stops or more.
What I find interesting about Ben's image though, is that the shaded areas appear to be very close to pure red, not orange-red. That would seem to imply that the light shining from above is contributing a strong yellow component, which does not really seem to be the case based on other lit objects in the image. There is also the situation of the red front of the engine on the left and the red horn, both of which are in the same light but neither of which turned yellow.
I'm still suspecting that the bumper of the subject engine used paint with some fluorescent or reflective properties that helped blow-out that portion of the image.
It is certainly an interesting case of saturation clipping.
p.3 #17 · 5d iii help: horrible over saturated reds.
I'm still suspecting that the bumper of the subject engine used paint with some fluorescent or reflective properties that helped blow-out that portion of the image.
Not to dismiss any of the previous discussion on reds but I think you might be onto something there. I was wondering about that. I have some red adhesive tape that I used on my boat trailer that looks just like red tape until you get light on it from certain angles or headlights at night.
I have seen some of my images that included road signs and licence plates end up a bit unusual.
p.3 #18 · 5d iii help: horrible over saturated reds.
Lotuselite wrote:
Not to dismiss any of the previous discussion on reds but I think you might be onto something there. I was wondering about that. I have some red adhesive tape that I used on my boat trailer that looks just like red tape until you get light on it from certain angles or headlights at night.
I have seen some of my images that included road signs and licence plates * end up a bit unusual. * and fire engines!
p.3 #19 · 5d iii help: horrible over saturated reds.
Photon wrote: * and fire engines!
and orange cones and people wearing glow in the dark orange
most of my DSLR shots of people taken during the 80's show extreme signs of such stuff going on various bits of clothing
edit: oops I didn't realize the topic changed to the different angle of view thing, that is certainly very real, but I doubt it applies to paint used on trains plus the normal and weird parts are viewed from the same angle
