GH1, meet the Stouffer 4110C dynamic range test

B

Barry_Green

Moderator
After the Billups/ToNi shooting session, I decided I needed a better way to test for definitive dynamic range.

Enter the Stouffer 41-step transmission wedge, which is a calibrated strip designed to transmit light in calibrated 1/3-stop increments. Each step on the wedge is 1/3-stop denser than the step that came before it. So, every three steps = 1 stop. To calculate a camera's total dynamic range, you shoot the chart in front of a bright light source, and examine the steps to see where you can clearly distinguish between them, and wherever you can't, you take note of that step as the point where the dynamic range of the chart has now exceeded the camera's ability to resolve it. Whatever the number is, divide by three and that's the # of stops of dynamic range you're looking at (so, if it discerns 22 but not 23, you'd stop at 22, divide by 3, and end up with 7 1/3 stops of total dynamic range).

Much more repeatable than my prior test, and certainly less subjective in evaluation.

So -- I ran the GH1 in every gamma mode, using contrast -2 to extend the dynamic range as far as possible. I also did one test with Intelligent Exposure On Max vs. Off, and one test where I ran the standard gamma curve through many of the ISO settings to see how dynamic range might change under higher ISOs.

In all these tests exposure was set so that the first step on the chart, step 1, the most translucent, was rendering on the waveform monitor as just barely darker than full clipped white.

So -- here are the results:

GH1-All-Gammas-100-Off.jpg



GH1-All-ISO.jpg



GH1-Standard-100-On-Off.jpg


Judge for yourself where you think the resolvable dynamic range ends.

As for my picks, I say that Smooth and Nostalgic definitely show more easily-discernible range than some of the others; at first glance it looks like "smooth" is 7 1/3, nostalgic is 7 2/3. But the fun thing is, the range really isn't changing any; the minimum and maximum are pretty much the same across all the gamma curves. They all pretty much max out at stop 25, if you pull them into photoshop and jack the exposure up so it's easier to tell. Vibrant seems short, it seems to actually cut off at about 23, and Nature's last discernible is 24, but Smooth and Standard hit 25. So that means we're looking at an absolute maximum of 8 1/3 stops in the best case, and 7 2/3 in the worst case, so not a lot of difference in maximum range. In just out-of-the-camera looks, yes, there's a big difference, Dynamic and Vibrant will give off a contrastier-appearing image, and Smooth and Nostalgic are flatter curves, but overall about the same amount of detail is still there and recoverable in post if need arises.

Higher ISOs do compromise the latitude a bit, at 400 ISO it loses maybe 1/3 stop as compared to at 100 ISO; at 800 and 1600 ISO I'd say it's about a stop down.

Intelligent Exposure (on level 3, maximum) definitely gives a boost to the shadows, but again I don't think it gives a true measurable increase in actual dynamic range; maybe 1/3 of a stop but that's about it.
 
Barry_Green said:
After the Billups/ToNi shooting session, I decided I needed a better way to test for definitive dynamic range.

Enter the Stouffer 41-step transmission wedge, which is a calibrated strip designed to transmit light in calibrated 1/3-stop increments. Each step on the wedge is 1/3-stop denser than the step that came before it. So, every three steps = 1 stop. To calculate a camera's total dynamic range, you shoot the chart in front of a bright light source, and examine the steps to see where you can clearly distinguish between them, and wherever you can't, you take note of that step as the point where the dynamic range of the chart has now exceeded the camera's ability to resolve it. Whatever the number is, divide by three and that's the # of stops of dynamic range you're looking at (so, if it discerns 22 but not 23, you'd stop at 22, divide by 3, and end up with 7 1/3 stops of total dynamic range).

Much more repeatable than my prior test, and certainly less subjective in evaluation.

So -- I ran the GH1 in every gamma mode, using contrast -2 to extend the dynamic range as far as possible. I also did one test with Intelligent Exposure On Max vs. Off, and one test where I ran the standard gamma curve through many of the ISO settings to see how dynamic range might change under higher ISOs.

In all these tests exposure was set so that the first step on the chart, step 1, the most translucent, was rendering on the waveform monitor as just barely darker than full clipped white.

So -- here are the results:

GH1-All-Gammas-100-Off.jpg



GH1-All-ISO.jpg



GH1-Standard-100-On-Off.jpg


Judge for yourself where you think the resolvable dynamic range ends.

As for my picks, I say that Smooth and Nostalgic definitely show more easily-discernible range than some of the others; at first glance it looks like "smooth" is 7 1/3, nostalgic is 7 2/3. But the fun thing is, the range really isn't changing any; the minimum and maximum are pretty much the same across all the gamma curves. They all pretty much max out at stop 25, if you pull them into photoshop and jack the exposure up so it's easier to tell. Vibrant seems short, it seems to actually cut off at about 23, and Nature's last discernible is 24, but Smooth and Standard hit 25. So that means we're looking at an absolute maximum of 8 1/3 stops in the best case, and 7 2/3 in the worst case, so not a lot of difference in maximum range. In just out-of-the-camera looks, yes, there's a big difference, Dynamic and Vibrant will give off a contrastier-appearing image, and Smooth and Nostalgic are flatter curves, but overall about the same amount of detail is still there and recoverable in post if need arises.

Higher ISOs do compromise the latitude a bit, at 400 ISO it loses maybe 1/3 stop as compared to at 100 ISO; at 800 and 1600 ISO I'd say it's about a stop down.

Intelligent Exposure (on level 3, maximum) definitely gives a boost to the shadows, but again I don't think it gives a true measurable increase in actual dynamic range; maybe 1/3 of a stop but that's about it.
Click to expand...



Thanks for the tests Barry! It's very much appreciated........


Barry, do you think the native ISO applies to the GH1 as well?


BTW, I have been shooting in "Nostalgic" for months now. When you -2 the contrast, it gives you that little bit extra in shadows with
low light environments. However I notice that it clips the top end a bit more then "smooth" from what I have seen....

.
.
 
Last edited:
I don't see the "native ISO" effect on the GH1; it seems to be pretty uniformly low in noise up until about 800, and then it starts to skyrocket. Each ISO bump gets a little noisier, but not like the Canon where 160 is actually quieter than 125 or 100.

Nostalgic seems to have a good shape to its latitude curve, but it tints the footage slightly reddish/sepia.
 
Thanks for going the extra mile on these tests Barry, a useful resource for the GH1.

I'm only looking at these charts on my laptop screen, but it does appear that I.E does lift a little in the shadows and takes away with the other hand in the highlights. I've rarely used this when filming, but when I have tested I've seen little to no difference either.
 
So the verdict is that "smooth" with -2 contrast produces the best results ? Wouldn't that make the video look washed out ?
 
Depends on what you mean by the "best" results. If you want the widest dynamic range, then yes, smooth with -2. And yes, wider dynamic range inherently means that an image looks more "washed out". Contrast makes images look sharper, and starker contrast means, inherently, less dynamic range. That's why you'll see the "dynamic" and "vibrant" gammas look punchier and more vibrant, but if you look at the chart you'll see that they appear to offer less overall dynamic range.
 
just a suggestion

just a suggestion

thanks for going through all the work to do this. i might suggeest you expose a bit lighter so that the usuable dynamic range falls more in the middle of the chart and does not start at one end. its all relative, so you would not need to expose exactly in the same spot for each shot, simply somewhere in the middle and measure the steps this way. for all we know, there could be more range available in the highlights though could not tell with these result. a little 'buffer' so to speak at each end (as dpreview does with the d-range tests) could be a bit more telling.

just a suggestion.. im no expert, so theorizing here
 
That wouldn't work with the chart though. The chart is designed to be exposed so that the very highest chip is just barely darker than pure white. That way you can quickly just look at the chip number and determine how many stops there are.

There is absolutely nothing in the highlights above this, because I exposed for that situation. There's nothing missing and there's nothing out there to recover.

I exposed way above and way below, and adjusted the exposure all throughout these shots, and then (using the waveform monitor in the NLE) picked the one that showed the ideal exposure, which is just where step 1 is barely below clipping.
 
Barry_Green said:
I exposed way above and way below, and adjusted the exposure all throughout these shots, and then (using the waveform monitor in the NLE) picked the one that showed the ideal exposure, which is just where step 1 is barely below clipping.
Click to expand...

Would you mind posting some of the shots that were exposed differently?
 
hey thanks! did not think it would be that grey at the far left end - guess you said its way overexposed though, so makes sense (no true black).

kind of interesting how shade #6 (which is just barely visible) in the overexposed image is lighter than #2 in all of the above, but a tad darker than #1.. jmo
 
Barry, Your tests are both interesting and accurate. I've found that the GH1 absolutely looks its best in the "smooth" mode, with the contrast at -2, sharpness at 0, color at 0, and NR(noise reduction) at -2. This yields the best looking image and tricks the camera's poor implementation of an AVCHD encoder into doing the best it can.

The setting described above are night and day better than the factory settings that most people never change. We're using a GH1 for "B" roll cutaways that integrate into an F900R show and (as long as you don't rapidly pan) it looks pretty good. Twice as good as our 7D test footage.
 
Glenn Przyborski said:
Barry, Your tests are both interesting and accurate. I've found that the GH1 absolutely looks its best in the "smooth" mode, with the contrast at -2, sharpness at 0, color at 0, and NR(noise reduction) at -2.
Click to expand...

Thanks alot Barry.

Glenn, very interesting. I have been shooting with the smooth preset since day one but I have both sharpness and color at -2 and I keep the NR at 0. I have been very satisfied with the results since I try to capture a flat digital negative with all my cameras. Any benefits to set NR at -2?
 
Prior to June of this year, I was pretty much ignorant about all aspects of DSLR video. Thanks to this forum and all the great contributors I've learned enough to now be shooting video with the GH1 that really has me excited about the capabilities of this camera.

Being now somewhat addicted (okay, totally) to this whole subject, I read virtually all posts and, from this, have pretty much settled on two presets: Standard and Smooth - depending on what I'm shooting.

I shoot a lot of nature footage and, yes, have struggled at times with the issues presented by green, finely detailed scenes, etc. Have even seen the codec struggle with dewdrops on the green leaves of a cornstalk where the slightest breeze caused the sun reflected in the dewdrops to shift the contrast in a very detailed way which was enough to muddy the image. However I've been able to overcome a lot of these issue by controlling DOF. But I digress.

Have been using the Standard mode with the following settings for the following reasons: Contrast at -2, Sharpness at +1, Saturation at 0, and NR at -2. The reason I have Sharpness at +1 is I shoot stuff with feathers and fur and had read that some detail can be lost with Sharpness set at 0. I also use Standard mode because I'd read that the Smooth mode tends to mute blues, and also due the the documented shifting problems with blue. In summary, I use Standard when there will be blue skies, etc. and/or lots of detail in feathers, etc.

Use the Smooth mode with the following settings for basically anything that doesn't call for Standard mode: Contrast at -2, Sharpness at 0, Saturation at 0, and NR at -2.

Have only recently finally gone with NR at -2 (was using 0) only because so many recommend it. And I've read that the sensor is good enough to not really need NR. Further explanations on the merits of NR at -2 would be appreciated.

In short, with the above settings I've found I can switch between Standard and Smooth and cover most situations. I'm sure there are individual tweaks that would further improve things, depending upon the exact scenario, but haven't felt comfortable enough with the camera yet to do much experimenting beyond these general settings.

Hopefully, this info may help someone or provide the basis for feedback to shed some further light on the subject.
 
The GH1 manual suggest either boosting NR in low light or lowering the other parameters. Dynamic range really seems to suffer in low light. Two questions:

1. What are people's best settings for low light?
2. The manual's suggestion presents the possibility that these settings are relative, that NR can either be boosted or lowered relative to other settings. Now the question, if you boost some of the other settings and lower NR to -2 will that produce the maximally reduced NR?
 
Last edited:
Anther related question:

From the many compressed online videos it appears to me that the WB in the GH1 is wildly off at times, especially in low light. Has anyone experimented with manually altering WB, not using a grey card but actually entering WB values, in low light and other extreme conditions? Does obessive monitoring of WB cure many of the perceived non-codec GH1 problems in extreme conditions? The reason that I include this question in a thread on dynamic range is that some of the results I see suggest, but certainly do not prove, that dynamic range is tied to WB in low light. But that could just be a perceptual effect that would not show up in a Stouffer test.
 
philiplipetz, I only use manual WB and sometimes I even tweak a certain setting i.e 3200K towards more magneta f.e. I usually shift between 5500K and 3200K, and at times 2700K.
 
philiplipetz said:
The GH1 manual suggest either boosting NR in low light or lowering the other parameters. Dynamic range really seems to suffer in low light. Two questions:

1. What are people's best settings for low light?
2. The manual's suggestion presents the possibility that these settings are relative, that NR can either be boosted or lowered relative to other settings. Now the question, if you boost some of the other settings and lower NR to 0 will that produce the maximally reduced NR?
Click to expand...

My guess about this is that it is for a general guide to those that use presets instead of the full manual feature on the camera. From what I can tell the dynamic range is at it's best from what Glenn Przyborski suggests.

Oh and Glenn, Pittsburgh eh? Rockin'! I went to film school there and it's one of my favorite cities. If only Knoxville would have such a rich undead heritage as Pittsburgh I wouldn't miss it so much. *sigh
 
tflak said:
Have been using the Standard mode with the following settings for the following reasons: Contrast at -2, Sharpness at +1, Saturation at 0, and NR at -2. The reason I have Sharpness at +1 is I shoot stuff with feathers and fur and had read that some detail can be lost with Sharpness set at 0. I also use Standard mode because I'd read that the Smooth mode tends to mute blues, and also due the the documented shifting problems with blue. In summary, I use Standard when there will be blue skies, etc. and/or lots of detail in feathers, etc.

Use the Smooth mode with the following settings for basically anything that doesn't call for Standard mode: Contrast at -2, Sharpness at 0, Saturation at 0, and NR at -2.

Have only recently finally gone with NR at -2 (was using 0) only because so many recommend it. And I've read that the sensor is good enough to not really need NR. Further explanations on the merits of NR at -2 would be appreciated.

Hopefully, this info may help someone or provide the basis for feedback to shed some further light on the subject.
Click to expand...

Here's my take on the GH-1's settings and why certain settings seem to work better than other settings... When you choose "smooth" you are technically raising the camera's gamma setting for the portion of the image between between about 10 to 50 units of video (out of 100 of course). This allows much more lower-light image detail to be resolved which would have been crushed or mushed into the shadows if you were using the "standard" setting. With the "smooth" setting, I have not noticed any loss of blue or any other specific colors.

Setting the "contrast" to -2 helps capture as much dynamic range into the image as possible. This is especially important on sunny, high-contrast days, or scenes with lots of highlights. Car headlights comes to mind.

In the GH-1, the "NR" or noise reduction setting is a big source of image problems. The GH-1 seems to minimize noise by reducing detail and literally blending together or averaging low-light areas of the scene. If you examine an area of dark brown bricks for example, you might see just a continuous color with no or very minimal detail. My gut felling is that this noise reduction method and/or algorithm was created for still images, not for video. (Sony & Panasonic's broadcast cameras don't use the GH-1's method of reducing noise.)

When you select "smooth", "contrast -2", "sharpness 0", and "NR -2" you fool the circuitry into minimizing the blurring or frame-to-frame averaging of low-light areas of your scene. Remember, Panasonic always keeps the NR on... "-2" is the lowest they'll allow you to select. I personally feel a little grain in a low-light image is far better than blended together areas of the picture.

Also, the "Nostalgic" setting ever further lifts low-light gamma, but it also seems to slightly "warm" the scene, which means lowering blue tones.
 
You must log in or register to reply here.
Back
Top