HS Hoops -- My Turn

KED · January 15, 2008

I'm really just posting this to stay in touch. Background: Just BEFORE my son was the last one cut from his varsity team (:cry ) I invested in an EF 50 f/1.4, which should have utility for my other kids' squash if they will ever let me shoot them (:cry ). Anyway, my big man decided to play JV, so yesterday I took my camera. I wanted to take on the indoor lighting challenge, and also to get a higher comfort level with the AI Servo on my Mk III ahead of the rapidly upcoming college lacrosse season.

Random observations: that "cycling light" phenomenon is amazing. I shot in RAW, and had stuff coming out in everything from bright yellow to dark blue, so thank goodness for RAW! And, uncomfortably, too early to tell about the AI Servo; it seems to have done OK, but there were too many variables including the DOF of the lens (shot everything at f/2.0). Finally, for it's vaunted high ISO capability (except in comparison to the D3 of course) I had to do an awful lot of noise reduction in PP for shots taken mostly at ISO 1000.

So these are just posted for fun, but with one question: in a shoot like this, when you are all done with PP and line up all your keepers in an array, is it a major gaffe if the WB doesn't necessarily look the same across the array? I find that some shots just look better, for whatever reason, with a different WB.

#1 Do not dis the shooter he's my kid!:

#2

#3 This is only up because it was about image #4 in a burst that I took deliberately to check the AI Servo, which in prior experience would have gone into meltdown by now (plus I kinda liked capturing the ball flat on the floor):

I know that better hoop drama comes from elevation and from dramatic contesting of shots. What can I tell you - it's JV, nobody has any ups, and the final was 78-25 so there wasn't much contesting going on.

C&C more than welcome, but mainly just saying "hey".

rwells · January 16, 2008

Hey KED,

When shooting in low light, I always try and make sure I'm not underexposing my shots. If you do, and your using a high ISO, your just begging for noise. Try to get the best exposure in camera, that's your best bet for as clean an image as you can get.

I personally like my pics WB to be as close to the same as I can get them. If your presenting several pics and the WB is different, the viewer has something to gauge against. If all are the same, and they are in the ballpark, it's a lot harder to pick up on it.

To me, your pics look about 1 stop underexposed. YMMV

Your original

I raised the exposure about a stop.
(I'll certainly pull this down if you wish)

You may well have different views on this, but you asked

KED · January 16, 2008

rwells wrote:

Hey KED,

To me, your pics look about 1 stop underexposed. YMMV

You may well have different views on this, but you asked

Good advice, as always, and the histograms agree with you about the underexposure -- but to my eye, at "proper" exposure per the histogram, the images were looking a bit too bright and washed out. There are other solutions to that, but I was getting impatient with PP.

I was trying to expose properly in-camera, but felt like I was maxed out aperture-wise (f/2.0) for sports, min'd out shutter speed-wise (400) for motion blur, so the only thing left was ISO. My logic was that you can, at least to some extent, correct for noise, but not at all for motion blur. Did I miss something there?

rwells · January 16, 2008

KED wrote:

Good advice, as always, and the histograms agree with you about the underexposure -- but to my eye, at "proper" exposure per the histogram, the images were looking a bit too bright and washed out. There are other solutions to that, but I was getting impatient with PP.

I was trying to expose properly in-camera, but felt like I was maxed out aperture-wise (f/2.0) for sports, min'd out shutter speed-wise (400) for motion blur, so the only thing left was ISO. My logic was that you can, at least to some extent, correct for noise, but not at all for motion blur. Did I miss something there?

KED,

I think your decisions were well founded, with the exception of the ISO choice.

I find that even when shooting the 30D I'm better off noise wise to up the ISO from 1000 to 1600 and be properly exposed than to underexpose even using a lower ISO. You'll get less noise this way.

Your also correct about getting washed out in PP, but that's my point, get it right in camera and you won't be pushing it in PP, and you'll have less noise to boot

It's a regular deal for me to shoot and sell 8x10 prints from ISO1600. If they were properly exposed, they look good. Now, I'm gonna' have to suspect that fancy 1DMkIII can out-do that with it's lens cap on

Give it a try, I'll bet ISO 1600 properly exposed on your camera is going to look really good...

KED · January 16, 2008

rwells wrote:

KED,

I think your decisions were well founded, with the exception of the ISO choice.

I find that even when shooting the 30D I'm better off noise wise to up the ISO from 1000 to 1600 and be properly exposed than to underexpose even using a lower ISO. You'll get less noise this way.

Your also correct about getting washed out in PP, but that's my point, get it right in camera and you won't be pushing it in PP, and you'll have less noise to boot

It's a regular deal for me to shoot and sell 8x10 prints from ISO1600. If they were properly exposed, they look good. Now, I'm gonna' have to suspect that fancy 1DMkIII can out-do that with it's lens cap on

Give it a try, I'll bet ISO 1600 properly exposed on your camera is going to look really good...

Wow . . . Wow! I had always thought that the discipline would have been to push ISO as the last resort. I am blown away and can't wait to try this. Won't have to wait too long -- should be able to try it next week! Maybe you've helped me discover some extra value in my fancy schmancy camera body to boot!

Anyway, 17 days till back outside for lacrosse in the snow, maybe even the sun -- THAT I should have a handle on by now!!! If not i am going to milk my press credential for all it's worth but leave the lens cap on!

rwells · January 16, 2008

KED wrote:

Wow . . . Wow! I had always thought that the discipline would have been to push ISO as the last resort.

Going to high ISO IS the last resort!

Trying to save underexposed high ISO images is like grave digging.
Your not going to like what you unearth...

jfriend · January 16, 2008

rwells wrote:

KED,
I think your decisions were well founded, with the exception of the ISO choice.

I find that even when shooting the 30D I'm better off noise wise to up the ISO from 1000 to 1600 and be properly exposed than to underexpose even using a lower ISO. You'll get less noise this way.

This is not a universally held belief when shooting RAW. About a year ago, there were some pretty detailed tests done on a Nikon camera that showed that when using good RAW software, ALL that really mattered for noise was how much exposure you applied (shutter speed and aperture) and it didn't matter whether you underexposed at a lower ISO or properly exposed at a higher ISO. The shutter speed and aperture are all that determines how many photons hit the sensor which is all that determines the signal-to-noise ratio on the sensor. The ISO does not affect this because that is a post-capture amplification step.

Underexposing and then making the image brighter in post processing is just amplifying the signal. Raising the ISO is just amplifying the same signal. It's different software doing it in one case vs. the other, but still you are starting with the same noise and ending with the same signal value. End result, you get about the same result if both pieces of software used in the comparison are decent.

If you are shooting JPEGs and using in-camera noise reduction, all bets are off because there is very different software involved in the two cases and a very different ordering of operations. Which is better in that case depends more on the software and skill to which it is applied.

rwells · January 16, 2008

jfriend wrote:

This is not a universally held belief when shooting RAW. About a year ago, there were some pretty detailed tests done on a Nikon camera that showed that when using good RAW software, ALL that really mattered for noise was how much exposure you applied (shutter speed and aperture) and it didn't matter whether you underexposed at a lower ISO or properly exposed at a higher ISO. The shutter speed and aperture are all that determines how many photons hit the sensor which is all that determines the signal-to-noise ratio on the sensor. The ISO does not affect this because that is a post-capture amplification step.

Underexposing and then making the image brighter in post processing is just amplifying the signal. Raising the ISO is just amplifying the same signal. It's different software doing it in one case vs. the other, but still you are starting with the same noise and ending with the same signal value. End result, you get about the same result if both pieces of software used in the comparison are decent.

If you are shooting JPEGs and using in-camera noise reduction, all bets are off because there is very different software involved in the two cases and a very different ordering of operations. Which is better in that case depends more on the software and skill to which it is applied.

I respectfully disagree...

Real world experience tells me this.

Also, read up on the many, good articles on "shooting to the right", as in to the right side of your histogram.

http://www.luminous-landscape.com/tutorials/expose-right.shtml

Here is a small quote from an excellent article from the above link.

"A 12 bit image is capable of recording 4,096 (2^12) discrete tonal values. One would think that therefore each F/Stop of the 5 stop range would be able to record some 850 (4096 / 5) of these steps. But, alas, this is not the case. The way that it really works is that the first (brightest) stop's worth of data contains 2048 of these steps — fully half of those available..........

...This realization carries with it a number of important lessons, the most important of them being that if you do not use the right-hand fifth of the histogram for recording some of your image you are in fact wasting fully half of the available encoding levels of your camera.."

Using a high ISO, take a picture with the histogram showing strong to the right, but not clipping of course. Now take the same shot with the histogram strong to the left, again not clipping. Now, bring the underexposed shot up to the correct exposure in post production.

I'll wait here until you come back with your results...

But, YMMV - I'm not holding a gun to anyone's head here

jfriend · January 17, 2008

rwells wrote:

I respectfully disagree...

Real world experience tells me this.

Also, read up on the many, good articles on "shooting to the right", as in to the right side of your histogram.

http://www.luminous-landscape.com/tutorials/expose-right.shtml

I know that the point I am making can be a bit counter-intuitive until you understand what really causes noise and how ISO really works. I thought the argument was a bit of heresy myself when I first encountered it. But, I think it's worth understanding.

I don't think you understood my earlier point exactly. Shooting to the right of the histogram is about getting the maximum exposure at a given ISO without clipping. That gets more photons on the sensor and does indeed help with noise. I'm not disagreeing with that. More light on the sensor will always reduce noise because if you hold the noise contributors constant and increase exposure with more light, you are improving the signal-to-noise ratio and thus the noise will be less visible. But raising the ISO is not more light on the sensor. It's just more amplification of the same signal in camera.

For example, a shot at 1/500th, f/2.8 and ISO 800 is exactly the same number of photons on the sensor as a shot at 1/500th, f/2.8 and ISO 1600 even though the histogram for the second will appear one stop to the right of the first. Both are putting the same number of photons on the sensor and giving you the same starting signal-to-noise ratio which is the key to visible noise. Even though the ISO 1600 shot will appear to have a histogram fully to the right, it won't have less noise than the ISO 800 shot with a centered histogram. The two will be very similar.

Even more to the point, you could get the same "to the right" histogram with 1/250th, f/2.8, ISO 800 and 1/500th, f/2.8, ISO 1600, but we know that the ISO 800 shot will have significantly less noise because it captured twice as many photons on the sensor (1/250th, f/2.8 vs. 1/500th, f/2.8). So, what matters the most for noise is how much light you capture, not where you set the ISO.

The analysis you quote from the luminous landscape article of the amount of stops of information contained in the right side of the histogram can only be used when staying at a constant ISO. When the ISO is held constant, then we're not in disagreement. Get as much light on the sensor as possible given your shooting situation (thus pushing the histogram as far right as possible). But raising the ISO just to push the histogram to the right isn't going to help you because you didn't add any light to the sensor and thus you didn't actually improve the signal-to-noise ratio at all.

As for your challenge of a shooting test, I've already done that test and I know the result. If you give the sensor more light, you get less visible noise. I'm not disagreeing with that conclusion at all. At a constant ISO, shooting to the right gets you less noise because you are capturing more photons. But this test doesn't work when comparing shots at different ISO.

Here's a test for you. Pick an exposure (shutter speed and aperture) that gets you a centered histogram at a raised ISO (say ISO 1600). Set your camera in manual exposure mode on those values and take a shot. Then, change the ISO to 3200 and take a shot. You will presumably get a histogram pushed to the right. Now adjust the second image down until it's the same brightness as the first one in post processing. You will have basically the same noise even though the ISO 3200 shot had a histogram pushed way to the right. This is because the sensor captured exactly the same number of photons in both scenarios. In the ISO 1600 case, you used the image as it came out of the camera. In the ISO 3200 case, you captured the same signal on the sensor, the camera amplified the signal an extra 2x in camera (over the ISO 1600 image) and you then did a 1/2x amplification in post processing. Net, net - same starting signal-to-noise, same net amplification of the signal and noise. The noise is equally visible in both case.

rwells · January 17, 2008

First off,

KED, I apologize for high-jacking your thread toward a camera/theory discussion :cry

But the last post has injected info that doesn't fit your scenario. IE: shooting sports in low light.

jfriend wrote:

I know that the point I am making can be a bit counter-intuitive until you understand what really causes noise and how ISO really works.

< I understand these things >

Shooting to the right of the histogram is about getting the maximum exposure at a given ISO without clipping. That gets more photons on the sensor and does indeed help with noise. I'm not disagreeing with that. More light on the sensor will always reduce noise because if you hold the noise contributors constant and increase exposure with more light, you are improving the signal-to-noise ratio and thus the noise will be less visible. But raising the ISO is not more light on the sensor. It's just more amplification of the same signal in camera.

< well, that should be understood, that's how changing ISO in a digital camera works -- But we'll get to this later >

Even though the ISO 1600 shot will appear to have a histogram fully to the right, it won't have less noise than the ISO 800 shot with a centered histogram. The two will be very similar.

So, what matters the most for noise is how much light you capture, not where you set the ISO.

< Ok, this is where you go off track - Yes, the noise will be about the same in this scenario, but with your ISO800 shot your picture is likely to be underexposed/too dark for use. Remember the original discussion here - shooting sports in a dimly lit gym, shutter speed up to keep blur to a minimum & aperture wide open, or close to it --- Yes, you CAN push the ISO 800 shot in post production, but now your really degrading the picture, not only letting the the noise through, but also posterization. (http://www.cambridgeincolour.com/tutorials/posterization.htm) At least with the ISO 1600 shot the picture is not underexposed and your not having to push it in post production.
Also, that's one less step to perform in post production which is very important to sports photography.>

The analysis you quote from the luminous landscape article of the amount of stops of information contained in the right side of the histogram can only be used when staying at a constant ISO. When the ISO is held constant, then we're not in disagreement. Get as much light on the sensor as possible given your shooting situation (thus pushing the histogram as far right as possible). But raising the ISO just to push the histogram to the right isn't going to help you because you didn't add any light to the sensor and thus you didn't actually improve the signal-to-noise ratio at all.

< Addressed above - again, you won't have to push the higher ISO shot in post production >

As for your challenge of a shooting test, I've already done that test and I know the result. If you give the sensor more light, you get less visible noise. I'm not disagreeing with that conclusion at all. At a constant ISO, shooting to the right gets you less noise because you are capturing more photons. But this test doesn't work when comparing shots at different ISO.

Here's a test for you. Pick an exposure (shutter speed and aperture) that gets you a centered histogram at a raised ISO (say ISO 1600). Set your camera in manual exposure mode on those values and take a shot. Then, change the ISO to 3200 and take a shot. You will presumably get a histogram pushed to the right. Now adjust the second image down until it's the same brightness as the first one in post processing. You will have basically the same noise even though the ISO 3200 shot had a histogram pushed way to the right. This is because the sensor captured exactly the same number of photons in both scenarios. In the ISO 1600 case, you used the image as it came out of the camera. In the ISO 3200 case, you captured the same signal on the sensor, the camera amplified the signal an extra 2x in camera (over the ISO 1600 image) and you then did a 1/2x amplification in post processing. Net, net - same starting signal-to-noise, same net amplification of the signal and noise. The noise is equally visible in both case.

< OK, I took the time & effort to conduct the test, but, I did it the other way around, like would happen in the real world. I took a shot of my kitchen counter at ISO 800, then only changed to ISO 1600 & took another shot. I cannot for the life of me, think of any reason to ever purposely shoot a higher ISO, then bring that exposure down to a lower ISO in post production. <img src="https://us.v-cdn.net/6029383/emoji/headscratch.gif" border="0" alt="" >

But, I can see the need to take a lower ISO shot and push the ISO up to get a correct exposure in post production. (that's what we are trying to keep from happening by choosing a higher ISO in the first place)

What this does is degrade the image worse than the correctly exposed higher ISO shot. Pushing an exposure in post production increases the visible noise and posterization occurs.

Again, why would someone do this in the real world?

So, I stand by my original statements to KED.>

Even an ISO800 shot that my camera handles well, when pushed in post production to ISO1600 shows posterization.

This is a screen shot where I have the ISO800 pushed to ISO1600 -- The lower overlaid picture is at ISO1600. Exactly the same crop size and magnification.

You can see posterization in the pushed ISO800 image (top image)

KED, I hope this helps, but again I apologize.

jfriend · January 17, 2008

rwells wrote:

Even an ISO800 shot that my camera handles well, when pushed in post production to ISO1600 shows posterization.

This is a screen shot where I have the ISO800 pushed to ISO1600 -- The lower overlaid picture is at ISO1600. Exactly the same crop size and magnification.

You can see posterization in the pushed ISO800 image (top image)

KED, I hope this helps, but again I apologize.

I'm sorry, but I don't see the effect you are trying to illustrate in the images you posted or understand the point you are making. Can you clarify what point you are making? If you left your meter on any automatic metering mode and just took two shots, one at ISO 800 and one at ISO 1600, the ISO 1600 shot is going to look worse because you only gave the sensor half as much light. Are you making that point or something else?

My point was that two shots with the same shutter speed and aperture settings such as a properly exposed ISO 800 and an "expose to the right" ISO 1600 shot will look nearly the same after the two shots have been equalized in post processing. In fact, the ISO 1600 shot might look a little worse. Thus, there's no advantage to raising the ISO in order to "expose to the right". On high ISO shots, you will get the best performance if you capture the most light you can (lowest shutter speed you can tolerate and largest aperture that will work), then set the ISO no higher than you need for a balanced histogram.

When shooting sports in a low light gym, I shoot RAW and I open my aperture up all the way. Pick the slowest shutter speed I think I can get away with and not have undesirable motion blur (for basketball, that's either 1/250 or 1/500 depending upon how desparate for light I am and what kind of shots I'm trying to capture). I then figure out what ISO will get my histogram in the range of a properly exposed histogram, but I don't try to push it to the right - I'm just looking to have no clipping on either end. If you end up at an ISO that you know just makes too much noise, then you either need to just deal with the noise, go with a slower shutter speed and try to capture shots that will work with that, acquire a lens with a larger aperture, use flash or consider a camera that has better high ISO performance.

rwells · January 17, 2008

jfriend wrote:

I'm sorry, but I don't see the effect you are trying to illustrate in the images you posted or understand the point you are making. Can you clarify what point you are making? quote]

John,

I'm not sure how much clearer I can make it. Maybe I'm just a bad communicator

My point in a nutshell:

I believe that a sport shooter is better off, if shooting in a dimly lit gym, using the widest aperture that you dare for your particular lens, the slowest shutter speed that you can live with to stop motion, then if you have to (and you usually do) bump up the ISO to get a properly exposed image in camera.

If you shoot at ISO800 (and the picture is underexposed) and you have to push the exposure in post, your not going to have as good an image as if you had shot the same picture at ISO1600 and was properly exposed (in this example) so you don't have to push the exposure in post.

Pushing an exposure in post WILL posterize an image. (you should be able to see that in the image that I posted)

OK, I'm done... this could go on forever.

jfriend · January 17, 2008

rwells wrote:

jfriend wrote:

I'm sorry, but I don't see the effect you are trying to illustrate in the images you posted or understand the point you are making. Can you clarify what point you are making? quote]

John,

I'm not sure how much clearer I can make it. Maybe I'm just a bad communicator

My point in a nutshell:

I believe that a sport shooter is better off, if shooting in a dimly lit gym, using the widest aperture that you dare for your particular lens, the slowest shutter speed that you can live with to stop motion, then if you have to (and you usually do) bump up the ISO to get a properly exposed image in camera.

If you shoot at ISO800 (and the picture is underexposed) and you have to push the exposure in post, your not going to have as good an image as if you had shot the same picture at ISO1600 and was properly exposed (in this example) so you don't have to push the exposure in post.

Pushing an exposure in post WILL posterize an image. (you should be able to see that in the image that I posted)

OK, I'm done... this could go on forever.

OK, I don't see any point of disagreement on your point. You haven't commented on my point, but that's fine. We can be done now.

I think I know what you meant, but usually "posterize" is not what happens when you push images in post processing. You get more noise which is not usually called posterization. In fact, noise tends to prevent posterization because it adds some randomness to the tonal and color variations.

KED · January 17, 2008

jfriend wrote:

rwells wrote:

OK, I don't see any point of disagreement on your point. You haven't commented on my point, but that's fine. We can be done now.

I think I know what you meant, but usually "posterize" is not what happens when you push images in post processing. You get more noise which is not usually called posterization. In fact, noise tends to prevent posterization because it adds some randomness to the tonal and color variations.

I LOVE this forum, and have learned from both of you since I have been a member. Right now my head is about to explode, but thankfully hoop was just for fun as i stated at the outset (who knew that a melee would ensue ). Honestly I don't know what to take away from this, I need to digest it more fully, but thank god I move outdoors for lacrosse in two weeks -- no more ISO/WB issues -- just the frozen tundra to deal with!!!

rwells · January 17, 2008

KED wrote:

...Honestly I don't know what to take away from this...

Yeah, I apologize for that, I didn't know it was going down that path either.

KED · January 18, 2008

rwells wrote:

Yeah, I apologize for that, I didn't know it was going down that path either.

You can stop apologizing now, amigo -- that was by no means a flamefest, only a healthy debate not unrelated to my original post or to our ensuing dialogue. I'm still trying to find a takeaway for my limited indoor/bad lighting needs, other than that this is an amazing group of really smart, experienced and generous people who've already made me a vastly better shooter.

HS Hoops -- My Turn

Comments