Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: dwswager on March 17, 2015, 01:40:29 pm
-
The answers I'm looking for do not involve buying a new camera, but I would like to help both myself and my friend.
Background
I shoot a lot of high school and club sports teams at night on very poorly lit fields. I have a friend that does the same and there are 2 other guys from local news organizations that I bump into regularly. When shooting at a new field that was very poorly lit, I looked around to find I was the only one shooting. When I asked my friend, he said that he could not get usable images in such light. I shoot a D810 and though the images have high noise and I wouldn't want to print them big, they are usable. I'm normally shooting Auto ISO up to 6400, 1/400th - 1/640th at f/2.8-f/4. None of us have $7000 1Dx or D3x or D4 cameras. My D810 is probably the most expensive one on the field and I'm the only Nikon shooter. On my D810 I sometimes underexpose to keep the ISO lower and shutter speed higher.
-
Hi,
Playing around with noise reduction and sharpening…
My ISO is glued to 50, leaving base ISO you give up quality, period.
Leaving base ISO is painful, but I have made decent quality A3-prints at 6400 ISO on my Sony Alpha 99.
Best regards
Erik
The answers I'm looking for do not involve buying a new camera, but I would like to help both myself and my friend.
Background
I shoot a lot of high school and club sports teams at night on very poorly lit fields. I have a friend that does the same and there are 2 other guys from local news organizations that I bump into regularly. When shooting at a new field that was very poorly lit, I looked around to find I was the only one shooting. When I asked my friend, he said that he could not get usable images in such light. I shoot a D810 and though the images have high noise and I wouldn't want to print them big, they are usable. I'm normally shooting Auto ISO up to 6400, 1/400th - 1/640th at f/2.8-f/4. None of us have $7000 1Dx or D3x or D4 cameras. My D810 is probably the most expensive one on the field and I'm the only Nikon shooter. On my D810 I sometimes underexpose to keep the ISO lower and shutter speed higher.
-
It all depends on what you grow up being used to, in terms of image quality at hi ISO's. I remember when. . .shooting tri-x and push processing to get ISO 3200. It gave you an image, which sometimes was even usable, with no shadow detail.
With the D800, you get quite "usable" stuff at 12,000! And of course you can even push it in post some too. Actually quite remarkable images to be had in the dark, which 10 years ago would never have happened.
There are things to be said, for not pushing ISO numbers, but if I had the choice of being tripod encumbered or not, I would chose leaving the tripod in the car, and pushing the ISO. Yep, sometimes in retrospect, I still kick myself for being lazy and not using the tripod on some shots. But my, how much quicker and easier it is to work without the tripod. I bet even Ansel would enjoy the convenience! ;-)
-
None of us have $7000 1Dx or D3x or D4 cameras.
if I am not mistaken Df has the same sensor as D4(s) and does not cost $7000, no ?
-
The answers I'm looking for do not involve buying a new camera, but I would like to help both myself and my friend.
Background
I shoot a lot of high school and club sports teams at night on very poorly lit fields. I have a friend that does the same and there are 2 other guys from local news organizations that I bump into regularly. When shooting at a new field that was very poorly lit, I looked around to find I was the only one shooting. When I asked my friend, he said that he could not get usable images in such light. I shoot a D810 and though the images have high noise and I wouldn't want to print them big, they are usable. I'm normally shooting Auto ISO up to 6400, 1/400th - 1/640th at f/2.8-f/4. None of us have $7000 1Dx or D3x or D4 cameras. My D810 is probably the most expensive one on the field and I'm the only Nikon shooter. On my D810 I sometimes underexpose to keep the ISO lower and shutter speed higher.
The D810 is not the hottest low light camera in the world, but it's no slouch, either.
http://blog.kasson.com/?p=8328
After you click on the link above, scroll down to the photographic dynamic range plot to comapre it to the a7II and a7R.
Here's a comparison with the D4 and a7S, two low light champs.
http://blog.kasson.com/?p=8280
You can see that it doesn't do too badly.
I advise underexposing in low light with the D810 by two or three stops and pushing the raw files in post. In-camera ISOs of over 1000 are achieved partially through digital gain, and you can do digital gain better yourself in post. Then apply nonlinear noise reduction in Lr or ACR, and res the image down to average out noise.
Jim
-
The D810 is not the hottest low light camera in the world, but it's no slouch, either.
http://blog.kasson.com/?p=8328
I advise underexposing in low light with the D810 by two or three stops and pushing the raw files in post. In-camera ISOs of over 1000 are achieved partially through digital gain, and you can do digital gain better yourself in post. Then apply nonlinear noise reduction in Lr or ACR, and res the image down to average out noise.
Thanks. Never tried under exposing that much. Usually just a stop. Will have to give it a try. I am still learning my way around Noiseware and ACR does OK. And while the controls are crude, the results for noise reduction from Allentechs Perfectly Clear 2 is actually pretty decent for down-sampled web images.
I actually get usable images without too much issue with the D810. It may not be the best low light camera on the market, but it is still very capable. And yes, leaving base ISO means giving up DR and low noise to get something else. This is the 1st camera I've been happy actually shooting at higher ISOs which is how I ended up shooting in Gyms and Night games in addition to outdoor day games.
My question was to help me, but more my friend who is using a Canon 7D and does not have the ability to buy something new. He has a wife! So I wanted to help him out if I could. While we both get paid from time to time for various endeavors, it is more a hobby and I still classify us as amateurs!
-
@Erik: Your cameras base iso is 100, setting it to Iso50 will cost you quite a bit of image quality, mainly dynamic range.
@dwswager: Your setup is almost the best money can buy, even changing to a D4s would only give you slightly more dynamic range but no improved noise. Guess you just have to accept the noise
-
Thanks. Never tried under exposing that much. Usually just a stop. Will have to give it a try. I am still learning my way around Noiseware and ACR does OK. And while the controls are crude, the results for noise reduction from Allentechs Perfectly Clear 2 is actually pretty decent for down-sampled web images.
I actually get usable images without too much issue with the D810. It may not be the best low light camera on the market, but it is still very capable. And yes, leaving base ISO means giving up DR and low noise to get something else. This is the 1st camera I've been happy actually shooting at higher ISOs which is how I ended up shooting in Gyms and Night games in addition to outdoor day games.
My question was to help me, but more my friend who is using a Canon 7D and does not have the ability to buy something new. He has a wife! So I wanted to help him out if I could. While we both get paid from time to time for various endeavors, it is more a hobby and I still classify us as amateurs!
The Canon 7D would not react well to underexposure. I believe that a combination of auto ISO and being observant on what ISO is chosen and shooting in manual setting shutter speed and aperture carefully with an eye on ISO is the way to go. Give up some DOF helps a lot. You don't mention his lenses. If they are f/5.6 he is already loosing two stops compared to an f/2.8 lens which I think should be default in light like you describe. Well you know this, of course :)
-
I advise underexposing in low light with the D810 by two or three stops and pushing the raw files in post. In-camera ISOs of over 1000 are achieved partially through digital gain, and you can do digital gain better yourself in post. Then apply nonlinear noise reduction in Lr or ACR, and res the image down to average out noise.
Jim, since the 810 is essentially isoless (I know you dislike the term) above ISO 800 or so, exposure can be achieved in the camera or in post with no penalty, except that the in camera method can result in blown highlights. What do you mean by underexposing by 2 or 3 stops and how would you go about achieving this? My own take would be that one would want to collect as many photons as possible, but would be limited by f/stop and shutter speed considerations for depth of field and freezing action. In this situation, one could use manual exposure and set the camera for the required shutter speed and f/stop and select the ISO such that the metered exposure value would be 2-3 stops underexposed according to the meter reading. One would then adjust exposure in post. Chimping would be dark under these conditions.
How would one use autoexposure under these conditions?
Regards,
Bill
-
Jim, since the 810 is essentially isoless (I know you dislike the term) above ISO 800 or so, exposure can be achieved in the camera or in post with no penalty, except that the in camera method can result in blown highlights. What do you mean by underexposing by 2 or 3 stops and how would you go about achieving this? My own take would be that one would want to collect as many photons as possible, but would be limited by f/stop and shutter speed considerations for depth of field and freezing action. In this situation, one could use manual exposure and set the camera for the required shutter speed and f/stop and select the ISO such that the metered exposure value would be 2-3 stops underexposed according to the meter reading. One would then adjust exposure in post. Chimping would be dark under these conditions.
How would one use autoexposure under these conditions?
Bill, I was trying to simplify without going into the whole ISOLess (and yes, I've knuckled under) exposure strategy thing. By underexposing two or three stops, I meant letting the histogram slide that far to the left from a conventional highlights just short of the right boundary histogram.
You're right. If you can't do ETTR at base ISO, you basically decide what f-stop and shutter speed you can live with, and that's what you set the camera at. But what do you do about ISO? My own most often used strategy is to set it to where I'll have to push it two or three stops in post. Why not more? Chimping, as you mentioned. Dark viewfinder, in mirrorless cameras. Twists in Lr developing. Things like that.
I often use autoexposure, too. Once I'm above base ISO, I set the mode to A, exposure compensation to -2 or -3, watch what shutter speeds the camera picks, and shoot away. This is my preferable way to do handheld panos, so that the camera can compensate for light level changes throughout the sequence, and AutoPano with do its exposure compensation thing when it stitches.
Jim
-
Thanks for the advice. Got my friend out tonight for a high school soccer game.
The guy with the 7D shoots at f/2.8 with a 70-200mm Canon lens. And the 7D has a 1.6x crop factor as an APS-C camera. I shot tonight with the D810 at f/2.8-f/3.5 with a AF-S Nikkor 70-200mm f/2.8 VRII w/ the camera set to a 1.2x crop (25MP).
Question: With 2 or 3 stops under exposure, are there some standard adjustments that are called for? Will the vibrance or saturation generally need bumped?
-
Question: With 2 or 3 stops under exposure, are there some standard adjustments that are called for? Will the vibrance or saturation generally need bumped?
No. See here: http://blog.kasson.com/?p=6574
And here http://blog.kasson.com/?p=6590
Jim
-
I did some night parade stuff with my D800s during Christmas and simply cranked up the ISO to 6400 and 12800. During pixel-peeping, they seem marginal, but with good processing of the raw files with ACR and certain images with Nik Define, I have a usable edit. When I travel to Serbia and the Balkans, I'm in a similar situation regarding dark monasteries, slow lenses and marginal conditions.
Normally for low light and this kind of work, I prefer shooting with my D3s which is nice, even at 12,800 and 25,000. I've gone higher when I've needed to do it.
I've been pushing the ISO for my kind of shooting now for several years.
The D700/D3 are fine up to 6400. The D3s gives me nice images at 12,000 and useable at 25,000. It's a matter of an image vs. no image in some cases. When I went with the D800 three years ago, at similar sizes (down-rezzed) to the D3s, it makes the grade. Pegged at 25,000 the D800 is a better image quality than the D700 by far.
For the most part, most images made pegging the ISO on the D800 are not necessarily useable at the full resolution for most images. However, up to say 11x17, all these cameras from the past several years are just phenomenal.
With good capture technique, good craft and good post processing, you can get good results. Part of it requires practice, part choosing just the best of the bad lighting situation.
I come from years of film where color above ISO 64 was hit-and-miss and Tri-X at 1600 was the state of the art. With the D800, I am fairly happy up to ISO 6400. Higher simply takes more post and the quality starts to suffer but is still doable.
Since the D800D810 has pixels to burn and most likely you are not doing 16x24 and larger images, simply keep pushing your camera and your craft and see how far you can go. I can tell you that of my last couple of museum shows, no one was the wiser on the photos shot at ISO 6400 nor even the 16x20s printed from an iPhone.
It all boils down to choosing the best images and then using good craft to make them shine!
-
... and res the image down to average out noise.
I don't think that down rezing is a preferred method to reduce visibility of noise. The "noise-supressing" properties of scaling can be had by using a plain linear filter (typically lowpass) without the aliasing-introducing sample-dropping inherent to image scaling. Using noise reduction software should be preferreable to both.
The noise-reducing properties of bilinear scaling should be well approximated by a "tent" lowpass filter of appropriate width. Lanczos2/3 kernels (and variations) might approximate more high-end scalers, but in the end I think it is about doing some smoothing of the upper band and managing the transition band without making the effective kernel so wide as to cause visible ringing.
If you have a high-rez image that you want to print (or display) on a high-rez medium, doing a "down-rez" operation actually means doing a "down-rez->up-rez" operation before the final image is rendered.
-h
-
I don't think that down rezing is a preferred method to reduce visibility of noise. The "noise-supressing" properties of scaling can be had by using a plain linear filter (typically lowpass) without the aliasing-introducing sample-dropping inherent to image scaling. Using noise reduction software should be preferreable to both.
The noise-reducing properties of bilinear scaling should be well approximated by a "tent" lowpass filter of appropriate width. Lanczos2/3 kernels (and variations) might approximate more high-end scalers, but in the end I think it is about doing some smoothing of the upper band and managing the transition band without making the effective kernel so wide as to cause visible ringing.
If you have a high-rez image that you want to print (or display) on a high-rez medium, doing a "down-rez" operation actually means doing a "down-rez->up-rez" operation before the final image is rendered.
-h
I don't disagree with any of this. I recommended nonlinear noise reduction before downsizing, and -- I'm sorry that I wasn't clear on this originally -- I assumed that, since the OP said he couldn't afford a D4, that the desired output was D4 size or smaller.
The D810 at high ISO is clearly worse than the D4 at 1:1. It is still worse when the images are viewed at the same resolution with good downsizing, but not much worse.
Jim
-
What would be a good starting point settings in ACR for a noisy night sport shot? I have never been able to do a whole lot in ACR to take care of the noise while not totally destroying detail.
I notice that when I use Noiseware, on a 16 bit image in Photoshop, it seems to do a much better job, but I'm looking for a good starting point to maximize the overall effectiveness.
Thanks for all the help guys! Love learning new things.
-
Thanks for the advice. Got my friend out tonight for a high school soccer game.
The guy with the 7D shoots at f/2.8 with a 70-200mm Canon lens. And the 7D has a 1.6x crop factor as an APS-C camera. I shot tonight with the D810 at f/2.8-f/3.5 with a AF-S Nikkor 70-200mm f/2.8 VRII w/ the camera set to a 1.2x crop (25MP).
Question: With 2 or 3 stops under exposure, are there some standard adjustments that are called for? Will the vibrance or saturation generally need bumped?
I have to underexpose 2 to 4 stops for my work. I'd luv to have your D810 noise problem. Look what I have to deal with.
3 to 4+ stops underexposed.
http://dewallenrld.tumblr.com/image/111042181561
http://dewallenrld.tumblr.com/image/111041717491
This one is as good as it gets for night noise with me
2 - 3 stops underexposed.
http://dewallenrld.tumblr.com/image/109786759681
all from
nsfw
http://dewallenrld.tumblr.com/
It is like shooting in the wet darkroom lit by the safelight only. Of course you can't use your monster cams for this work. I use 12 to 16 mp. All my shots are candids. If you guys tried to get close like I do with your bloated mega pixel cams the girl will come out and throw bleach or ammonia on you. Or she will signal the pimp or enforcers to kick your ass and throw you cam in the canal. That is why most of what you see in the red light district are shot with a zoom from across the canal.
My advice is if you need to make bigger prints make a master prints and you can scan it and enlarge from the scan. PP only does so much before the image starts to fall apart. For my social doc work this noise level is acceptable. For everyday stuff this noise level would not be useable for my taste.
-
in some situation, natural light really not giving you lots of options. Try to use off camera speedlight.
-
I have to underexpose 2 to 4 stops for my work. I'd luv to have your D810 noise problem. Look what I have to deal with.
3 to 4+ stops underexposed.
http://dewallenrld.tumblr.com/image/111042181561
http://dewallenrld.tumblr.com/image/111041717491
This one is as good as it gets for night noise with me
2 - 3 stops underexposed.
http://dewallenrld.tumblr.com/image/109786759681
Nice! If I was shooting this it would probably be with the 16-35mm f/4 VR and then instead of how to keep the shutter speed up, it would be how low can I go not to blur movement!
-
On my D810 I sometimes underexpose to keep the ISO lower and shutter speed higher.
In your D810 doing this is not a bad thing, or at least it is not worse than pushing ISO to achieve higher RAW values. But if any of your friends have Canon cameras, or Nikon cameras with Nikon (not Sony) sensors, underexposing at low ISO is the worst they can do. They'll get better results by pushing ISO for the same exposure (aperture/shutter) because on Canon/Nikon sensors captured DR is improved a lot by pushing ISO (as long as there are no blown highlights of course, which is the case in the underexposure conditions you mention).
2 shots at the same exposure (aperture/shutter), just changing ISO on a Canon-type sensor:
(http://www.guillermoluijk.com/article/iso/versus.jpg)
-
In your D810 doing this is not a bad thing, or at least it is not worse than pushing ISO to achieve higher RAW values. But if any of your friends have Canon cameras, or Nikon cameras with Nikon (not Sony) sensors, underexposing at low ISO is the worst they can do. They'll get better results by pushing ISO for the same exposure (aperture/shutter) because on Canon/Nikon sensors captured DR is improved a lot by pushing ISO (as long as there are no blown highlights of course, which is the case in the underexposure conditions you mention).
However, that benefit for Canon cameras stops approx. around ISO 800 - 1600. Any subsequently higher ISO setting may even reduce (or at least not improve) the Signal to Noise ratio.
2 shots at the same exposure (aperture/shutter), just changing ISO on a Canon-type sensor:
Yes, but now compare equal exposure at ISO 800 and 12800 ...
Cheers,
Bart
-
What would be a good starting point settings in ACR for a noisy night sport shot? I have never been able to do a whole lot in ACR to take care of the noise while not totally destroying detail.
Thanks for all the help guys! Love learning new things.
My experience with ACR/LR noise reduction is to treat the subject and the background separately. Usually I will deal with the background using the global detail controls. Obviously less sharpening, more masking and more luminance and color noise reduction reduces noise the most. You'll have to experiment on your images. When you have the background where you want it use an adjustment brush to apply negative noise reduction on the subject as necessary.
-
In your D810 doing this is not a bad thing, or at least it is not worse than pushing ISO to achieve higher RAW values. But if any of your friends have Canon cameras, or Nikon cameras with Nikon (not Sony) sensors, underexposing at low ISO is the worst they can do. They'll get better results by pushing ISO for the same exposure (aperture/shutter) because on Canon/Nikon sensors captured DR is improved a lot by pushing ISO (as long as there are no blown highlights of course, which is the case in the underexposure conditions you mention).
2 shots at the same exposure (aperture/shutter), just changing ISO on a Canon-type sensor:
(http://www.guillermoluijk.com/article/iso/versus.jpg)
Was the 1600 ISO one exposed right and pulled down in post or was the ISO 100 one under exposed and pull up?
My issue is I'm pretty locked in for shutter speed and aperture. So the assumpition is I'm shooting 1/640th to stop action and f/2.8 as much light as I can. Hence the choice is let ISO ride as far as possible OR underexpose by 2-3 stops retarding the ISO rise by those 2-3 stops. I used to under expose by 1 stop but I think I now have the hang of pulling 2-3stops.
-
My experience with ACR/LR noise reduction is to treat the subject and the background separately. Usually I will deal with the background using the global detail controls. Obviously less sharpening, more masking and more luminance and color noise reduction reduces noise the most. You'll have to experiment on your images. When you have the background where you want it use an adjustment brush to apply negative noise reduction on the subject as necessary.
Thanks. That is the type of real world advice that helps. I actually hadn't thought about negative noise reduction brush adjustments!
-
BTW, Here is a sample from a game a couple nights ago so you can see where I'm at. D810 shot at -1.7EV exposure compensation at ISO 4500 Nikon NEF. Adjusted in ACR and then downsampled and sharpened in Photoshop.
-
BTW, Here is a sample from a game a couple nights ago so you can see where I'm at. D810 shot at -1.7EV exposure compensation at ISO 4500 Nikon NEF. Adjusted in ACR and then downsampled and sharpened in Photoshop.
Looks pretty good to me!
-
Looks pretty good to me!
+1
I agree with Jim, the results look pretty good. Looking at the exif, I see the exposure was -1.67 EV on the camera and +1.75 in ACR and the ISO was 4500 with manual exposure. For my own benefit (and possibly others), I would be interested in how you achieved these results, that is how you metered and set the ISO and arrived at the shutter speed and aperture setting. Presumably as Jim described above?
Thanks,
Bill
-
BTW, Here is a sample from a game a couple nights ago so you can see where I'm at. D810 shot at -1.7EV exposure compensation at ISO 4500 Nikon NEF. Adjusted in ACR and then downsampled and sharpened in Photoshop.
Looks okay to me. The skin has that noise reduction plastic look. I think you can improve that with a little selective negative noise reduction if you are using ACR/LR.
-
+1
I agree with Jim, the results look pretty good. Looking at the exif, I see the exposure was -1.67 EV on the camera and +1.75 in ACR and the ISO was 4500 with manual exposure. For my own benefit (and possibly others), I would be interested in how you achieved these results, that is how you metered and set the ISO and arrived at the shutter speed and aperture setting. Presumably as Jim described above?
Thanks,
Bill
That's right -1 2/3 stops under. Was still afraid to go lower. Will try this weekend.
D810 w/ AF-S Nikkor 70-200mm f/2.8
Auto ISO 64-600
Manual Mode
-1 2/3 stops Exposure Compensation
Center Weighted Metering
Shutter Speed 1/640th (manually set)
Aperture f/3.5 (manually set)
ISO 4500 (Auto, combination of the shutter/aperture and exposure compensation)
Image Area set to 1.2x (25MP)
Auto Focusing
White Balance pre set using an Expodisc (White balance is my hardest problem and just changing the angle gives different results. I don't have a dual profile for this field which is usually what I do for night games. I get image in daylight and then at night at a field and use that as the starting profile.)
In ACR, I used the Flat profile as the starting point.
-
I advise underexposing in low light with the D810 by two or three stops and pushing the raw files in post. In-camera ISOs of over 1000 are achieved partially through digital gain, and you can do digital gain better yourself in post. Then apply nonlinear noise reduction in Lr or ACR, and res the image down to average out noise.
This test was sort of asked before, but I did not really see an answer: given that one is constrained to a certain aperture and shutter speed, I can see setting the "ISO" (in-camera gain setting) a bit below what would given an on-meter exposure in order to protect highlights from amplifier clipping (not photo-site blow-outs), but I am unclear why one would use as much as two or three stops. Is there evidence of a benefit in going beyond about one stop of under amplification (so-called "underexposure" but actually exposer level is fixed by aperture and shutter speed choice)? There would need to be some IQ benefit, to offset the less usable in-camera reviews and such.
Also, I understand that in many cameras, increasing EI beyond some level has no effect on the analog gain, and simply sets a flag in the raw file about suggesting a bit shift in raw conversion. In that case, I would expect that those very high EI setting would not have a problem with blown highlight in the photo-sites themselves, and analog gain would still be leaving the image "dim" in terms of raw levels, so no risk of amplifier clipping either. Or do some cameras apply bit-shifting "digital amplification" when producing raw files at these high EI settings? That would seem foolish.
(It would also seem foolish for a camera maker to use huge amounts of analog gain; once analog gain lifts the sensor noise floor clearly above the noise floor downstream from there, digital gain seems both better and cheaper. Does anyone know the limits of analog gain used in the various Sony/Nikon sensors?)
-
Also, I understand that in many cameras, increasing EI beyond some level has no effect on the analog gain, and simply sets a flag in the raw file about suggesting a bit shift in raw conversion. In that case, I would expect that those very high EI setting would not have a problem with blown highlight in the photo-sites themselves, and analog gain would still be leaving the image "dim" in terms of raw levels, so no risk of amplifier clipping either. Or do some cameras apply bit-shifting "digital amplification" when producing raw files at these high EI settings? That would seem foolish.
I am told that some Sigmas work that way. However, in my experience with Sonys (a7 a7S, a7R, a7II, RX-1, NEX7) and Nikons,(D3, D3s, D4, D800E, D810) they actually increase the values in the raw files, and then you can get clipping.
Jim
-
I am told that some Sigmas work that way. However, in my experience with Sonys (a7 a7S, a7R, a7II, RX-1, NEX7) and Nikons,(D3, D3s, D4, D800E, D810) they actually increase the values in the raw files, and then you can get clipping.
Jim
Pity; even my ancient Olympus E-1 did it that way: naive raw convertors that did not know about the flags produced underexposed conversions at EI 3200 and 1600, since the processing to raw was effectively capped at 800 (or was it 1600? It has been a while).
I will try to test my newer Olympus bodies -- unless someone like Guillermo L. has already done so and gives us the answer!
-
That's right -1 2/3 stops under. Was still afraid to go lower. Will try this weekend.
White Balance pre set using an Expodisc (White balance is my hardest problem and just changing the angle gives different results. I don't have a dual profile for this field which is usually what I do for night games. I get image in daylight and then at night at a field and use that as the starting profile.)
In ACR, I used the Flat profile as the starting point.
Thanks for the detailed reply. That's pretty well what I would do. I'ts been awhile since I shot sports under the lights since my son has graduated from high school. I presume the lighting is metal halide without a very high color rendering index. What I did was to take a reading under the lights from a Whibal and use this as a preset for WB. It worked reasonably well. If you have the x-rite ColorChecker Passport, it might be worthwhile to use this to make a custom profile.
Regards,
Bill
-
For those that advise under exposure, with the D810 is there a crossover. I'm trying do figure an overall underexposure versus ISO strategy.
What I mean is at what ISO should I start underexposure? Is it better to shoot ISO 1600 with 2 stops underexposure or ISO 3200 with 1 stop under or ISO 6400 with 0 under exposure? What I want to do is build a general set of guidelines on when to let ISO ride and when to use underexposure.
-
For those that advise under exposure, with the D810 is there a crossover. I'm trying do figure an overall underexposure versus ISO strategy.
What I mean is at what ISO should I start underexposure? Is it better to shoot ISO 1600 with 2 stops underexposure or ISO 3200 with 1 stop under or ISO 6400 with 0 under exposure? What I want to do is build a general set of guidelines on when to let ISO ride and when to use underexposure.
You might want to take a look at these curves:
http://blog.kasson.com/?p=8230
Read the preceding couple of posts for background.
You can see that the answer to your question depends on which shadow zone is most important to you.
If you care most about the deepest possible shadows, then read noise vs ISO is important to you:
http://blog.kasson.com/?p=6552
Jim
-
You might want to take a look at these curves:
http://blog.kasson.com/?p=8230
Read the preceding couple of posts for background.
You can see that the answer to your question depends on which shadow zone is most important to you.
If you care most about the deepest possible shadows, then read noise vs ISO is important to you:
http://blog.kasson.com/?p=6552
Jim
Thanks. Not really looking for the deepest shadows, but minimizing noise.
-
Thanks. Not really looking for the deepest shadows, but minimizing noise.
Then maybe you should use the mean = 0.04 curve (4 1/2 stops down from full scale). Then the answer to your question is ISO 200. But I don't like to push more than 4 stops in post, because of twists and other usability problems.
Jim
-
From all of these curves and discussion, I see no drawback to letting ISO go as high as warranted by the light level and constrained exposure setting. There may be no S/N or dynamic range benefit beyond a certain point, but I see no penalty either. Just hold back enough to avoid highlight clipping. Am I missing something?
-
From all of these curves and discussion, I see no drawback to letting ISO go as high as warranted by the light level and constrained exposure setting. There may be no S/N or dynamic range benefit beyond a certain point, but I see no penalty either. Just hold back enough to avoid highlight clipping. Am I missing something?
Not really. You won't get the film-like shoulder of the PV 2012 Exposure control, but you can do that manually.
If you get into the ISO region where the camera applies gain digitally, you'll be able to do it with greater precision (in the computer science sense of the word) in post.
Jim
-
From all of these curves and discussion, I see no drawback to letting ISO go as high as warranted by the light level and constrained exposure setting. There may be no S/N or dynamic range benefit beyond a certain point, but I see no penalty either. Just hold back enough to avoid highlight clipping. Am I missing something?
The exposure system of most cameras does not guarantee against highlight clipping afaik. The in-camera histograms and "blinkies" are not robust indicators of raw file clipping. Thus there is some risk of clipping highlights if you increase ISO (either automatically or manually), while there appears to be little to no raw-file IQ benefit beyond some threshold (perhaps ISO 100 for some cameras, perhaps ISO1600 or so for Canon cameras).
Of course, there are operational benefits to using in-camera ISO. You get camera previews and jpegs that make sense. You get raw files that are well aligned with what many raw file developers are optimized for.
-h
-
That's right -1 2/3 stops under. Was still afraid to go lower. Will try this weekend.
D810 w/ AF-S Nikkor 70-200mm f/2.8
Auto ISO 64-600
Manual Mode
-1 2/3 stops Exposure Compensation
Center Weighted Metering
Shutter Speed 1/640th (manually set)
Aperture f/3.5 (manually set)
ISO 4500 (Auto, combination of the shutter/aperture and exposure compensation)
Image Area set to 1.2x (25MP)
Nice picture. May I ask why EC was set to -1 2/3 with Exposure fixed manually? That forces a brightness correction on every image in post. It would seem more efficient to leave EC at 0EV and simply set Max Auto ISO so that the brightest desirable highlights are safely below clipping (-1 2/3 stops?). And following what Jim and -h have said, in any case no higher than ISO3200 for the D810.
Jack
-
The exposure system of most cameras does not guarantee against highlight clipping afaik. The in-camera histograms and "blinkies" are not robust indicators of raw file clipping. Thus there is some risk of clipping highlights if you increase ISO ...
This concern is independent of ISO setting. Even at low ISO, with plenty of light, you have to expose to avoid clipping. Just keep doing that at high ISO.
-
This concern is independent of ISO setting. Even at low ISO, with plenty of light, you have to expose to avoid clipping. Just keep doing that at high ISO.
Well, but by reducing ISO you can (practically) remove the clipping concern altogether. Choose aperture and exposure time for aesthetical reasons and fire away (some caution is needed of course to avoid too much underexposure).
-k
-
Nice picture. May I ask why EC was set to -1 2/3 with Exposure fixed manually? That forces a brightness correction on every image in post. It would seem more efficient to leave EC at 0EV and simply set Max Auto ISO so that the brightest desirable highlights are safely below clipping (-1 2/3 stops?). And following what Jim and -h have said, in any case no higher than ISO3200 for the D810.
Jack
Setting EC in Manual Mode on the D810 means the camera will adjust exposure computations solely with ISO. Hence, I set negative EC to keep ISO down. Typically I won't let it get higher than 6400 and usually is in the range of 1250-4000. So if I set -2 stops EC and the shutter is fixed at 1/640 and aperture at f/2.8 and ISO reads 3200, then without the EC, the camera would set the ISO to 128,000. BTW, not many cameras worked this way in Auto ISO mode in the past and was a Rant by Michael.
-
This concern is independent of ISO setting. Even at low ISO, with plenty of light, you have to expose to avoid clipping. Just keep doing that at high ISO.
What you say is true, but the risk/reward calculus is different if you're at base ISO and if you're not. At base ISO, ETTR actually buys you something: increased photon noise SNR, and read noise SNR, too, although at base ISO that is usually not the problem. Since there's a reward, it's worthwhile to take a few risks with the highlights. Now, say, for some good reasons you can't put that much light on the sensor, so you'll have to stop down from that ideal exposure, or go for a shorter shutter-open time. In an ISO-less camera, there's little reward for artificially pushing the in-camera histogram to the right, and the risk of highlight clipping is still the same if you do.
Jim
-
So if I set -2 stops EC and the shutter is fixed at 1/640 and aperture at f/2.8 and ISO reads 3200, then without the EC, the camera would set the ISO to 12,800.
Right, nothing wrong with that. I am merely suggesting that if desirable highlights are appropriately away from clipping with those settings and scene, you could achieve the same objective by leaving EC=0 and setting Maximum ISO at 3200. This would for instance save someone like me innumerable underexposed shots the next day out, when the camera gets quickly pulled out , switched to A mode and a few shots fired for a 'moment'... only to discover later that they were all captured with EC = -1 2/3 :)
Cheers,
Jack
-
Right, nothing wrong with that. I am merely suggesting that if desirable highlights are appropriately away from clipping with those settings and scene, you could achieve the same objective by leaving EC=0 and setting Maximum ISO at 3200. This would for instance save someone like me innumerable underexposed shots the next day out, when the camera gets quickly pulled out , switched to A mode and a few shots fired for a 'moment'... only to discover later that they were all captured with EC = -1 2/3 :)
Cheers,
Jack
1. Yes, it is almost criminal that modern cameras, where everything is set electronically, do not have a system to save and recall the entire state of the camera. The shooting banks on Nikon pro cameras are almost worse than nothing because it is easy to change something and forget that it was changed during shooting and then go back to that bank thinking you are back to the base state you want.
2. I'm actually trying to figure out what is best. Do I want 0EV compensation and ISO 3200 or -2EV compensation and ISO 800, for example? I seem to be hearing from Jim Kasson that it is worth staying under ISO 1000 which is where the D810 starts applying digital gain when possible.
Most games I shoot start in late afternoon where assuming 1/640 and f/2.8, for example, the ISO would normally range between 320-1250. By the end of the game, it is normally ranging between about 1600-6400. That is due to loss of the ambient light filling in the stadium lights. Of course, the light color shift is the other issue.
-
I seem to be hearing from Jim Kasson that it is worth staying under ISO 1000 which is where the D810 starts applying digital gain when possible.
That's a good point. I guess it comes down to whether one believes that the D810's input referred read noise drop at ISO 3200 (http://home.comcast.net/~NikonD70/Charts/RN_e.htm#D810_14) is real or subliminal NR. I tend to think it's real mainly because it would be unusual for Nikon to be doing NR in the green channel at ISO 3200 (http://www.dpreview.com/forums/post/54802684) (never seen it, at least under my watch) - but I don't have a D810 and I don't know of anyone who tested for NR there.
Jack
-
Here is an example of a 2 to 5 stop push I did with a Fuji X.
https://danielteolijr.wordpress.com/2015/04/13/an-example-of-push-processing-a-digital-image/
-
I seem to be hearing from Jim Kasson that it is worth staying under ISO 1000 which is where the D810 starts applying digital gain when possible.
It seems to me that all this agonizing would be avoided if a camera were to
(a) only ever apply as much analog gain as is beneficial to improving shadow noise (maybe only to about ISO 800 with many modern sensors?), and
(b) applying digital gain only during raw-JPEG conversion, indicating this in raw files only with a flag rather than actually increasing the numerical levels recorded in the raw file.
That would seem to offer the best of both worlds for low light high exposure index photography:
- minimizing the effects of read noise through adequate analog gain,
- live view and in-camera reviews displaying at the intended level placement, rather than appearing too dark, and
- avoiding any highlight damage due to excessive amplification, either analog or digital.
A question: when cameras use digital rather than analog gain beyond a certain ISO speed setting, which ones apply this to the raw levels (dumb!), and which do it as I propose above, only in conversion (the safer approach, used even by my ancient CCD-powered Olympus E-1 for its highest ISO speed settings.)
-
A question: when cameras use digital rather than analog gain beyond a certain ISO speed setting, which ones apply this to the raw levels (dumb!), and which do it as I propose above, only in conversion
As far as I know, only the Fujifilm X-100 uses your (and mine) preferred approach:
(http://www.guillermoluijk.com/article/dxomark/isox100_large.gif)
-
1. Yes, it is almost criminal that modern cameras, where everything is set electronically, do not have a system to save and recall the entire state of the camera. The shooting banks on Nikon pro cameras are almost worse than nothing because it is easy to change something and forget that it was changed during shooting and then go back to that bank thinking you are back to the base state you want.
On my Canon camera, if you are using one of the Custom Settings, it reverts to the predefined setting if it goes into sleep mode or you turn the camera off. I can sort of understand why Nikon have taken the approach they did but I prefer the Canon approach.
-
As far as I know, only the Fujifilm X-100 uses your (and mine) preferred approach:
I was under the impression that some MFDB also used this approach:
http://www.dxomark.com/Cameras/Hasselblad/H3DII-50---Measurements
I find it really strange that manufacturers does not follow simple engineering best-practice. Especially since raw formats are proprietary and they get to define both the encoding and decoding behaviour (3rd party will have to reverse-engineer anyways). What are their assumptions about their customers?
Perhaps an ETTR "preprocessor" for Lightroom is in order? Parse the raw file. Apply the specified EC to the raw values directly (taking care of bias) and round the result. If overflow occurs, split into 2 files (mimic exposure bracketing). Import into Lightroom.
For my 7D, this is not such a big problem. I should (if possible) bump ISO to 1600/3200 anyways to reduce noise (Canon are not "ISO-less"). I don't care much for under-exposing a lot beyond that point now matter if the gain is applied in-camera or in raw editor. Further, I generate my own camera profiles, and I believe that they are without twisting.
-h
-
I was under the impression that some MFDB also used this approach:
http://www.dxomark.com/Cameras/Hasselblad/H3DII-50---Measurements
That approach of fixed gain (a fixed ratio of numerical raw levels per photo-electron) makes particular sense for MF backs with CCD's, where (a) any gain must be applied relatively late (after transport to the corner of the sensor at least) so that increased gain is of little help is reducing dark noise, and (b) the ADCs are usually good enough ("16-bit"?) that the sensor output can be reliably handled with a fixed gain regardless of the exposure index in use. (That is, a single amplification level can simultaneously handle full wells without clipping while having ADC discretization noise well below the noise floor in the signal coming from the sensor.)
And yet AFAIK, not all MFD backs do it that way: maybe engineering bet practice clashes with the expectations of some users and hence leads to push-back from the marketing division. Remember all the talk that any such failure to amplify the raw levels up to the point of leaving bare minimum acceptable headroom makes higher ISO settings "fake", as still propagated by DXO's ill-informed declaration that the H3D50-II actually has the same "ISO" of 50 at all its ISO exposure index settings. That is, adopting engineering best practice can get you accused of dishonesty and/or technical inferiority.
-
Remember all the talk that any such failure to amplify the raw levels up to the point of leaving bare minimum acceptable headroom makes higher ISO settings "fake", as still propagated by DXO's ill-informed declaration that the H3D50-II actually has the same "ISO" of 50 at all its ISO exposure index settings. That is, adopting engineering best practice can get you accused of dishonesty and/or technical inferiority.
The same happens when a particular camera has specially low real ISO values vs declared ISO (which should be the trend while sensors improve their dynamic range), like most of the Olympus for instance: ISO200 meters for ISO200 but amplifies the RAW data by an equivalent ISO100. People say: hey! ISO100?!? that is cheating to get lower noise!!! funily a deeper understanding of what's going on will show that low effective ISO's usually mean the opposite, a challenge to keep SNR at good ratios because they basically produce RAW underexposure. The good news is helping to prevent disgusting blown highlights by compressing the right end of the camera JPEG histogram in a more pleasant and filmic way.
Regards
-
The good news is helping to prevent disgusting blown highlights by compressing the right end of the camera JPEG histogram in a more pleasant and filmic way.
And, for raw shooters, aided and abetted by the PV 2012 Exposure slider in Lr, which introduces a "shoulder" like film.
Jim
-
On my Canon camera, if you are using one of the Custom Settings, it reverts to the predefined setting if it goes into sleep mode or you turn the camera off. I can sort of understand why Nikon have taken the approach they did but I prefer the Canon approach.
I like Nikon's approach to that situation in that the camera holds the set values. There is a big difference between the camera going to sleep or being turned off and having the ability to select a set of predetermined settings.
How every camera should work:
1. Set the way you want say for Night Sports
2. Save the settings, preferably giving it a unique menu name.
3. You select the menu item Night Sports and the camera returns to the state when you saved the settings.
4. During shooting you decide to back Auto ISO down 2 stops and lower the minimum shutter speed.
5. The camera is now in a new state and when it goes to sleep or is turned off it should come back on in the same state as sleep or off.
6. But, when you select Night Sports again, the camera goes back to the original Night Sports settings you actually saved, not changed during shooting.
7. Of course, you are always able to change the settings of Night Sports to whatever you prefer.
I don't want a saved mode where changes to the settings while in that mode become the new settings for the mode, I want a set of home base states[/b] that I can quickly get back to. But the camera should only change back to the home base state when I tell it, not on it's own because that was the last state I set. If the camera allowed you to set 10 states and give them menu names, then it would require 11 states to be saved. The 10 you set, and the current state of the camera at the time it is turned off or goes to sleep. It would always power on in the last current state, but you could always choose a different state.
-
I don't want a saved mode where changes to the settings while in that mode become the new settings for the mode, I want a set of home base states that I can quickly get back to.
Agreed 100%. Unfortunately, Nikon's approach with the custom banks does not work that way. In the mean time, I follow Joe McNally's suggestion of "resetting" the camera to your default settings before every session / day.