X-E2 DR400 limits test

Started Mar 18, 2014 | Discussions thread
Ysarex
OP Ysarex Senior Member • Posts: 1,430
Re: X-E2 DR400 limits test
1

nixda wrote:

Ysarex wrote:

nixda wrote:

Ysarex wrote:

nixda wrote:

You were complaining that there is a lot of wasted space in the raw data at the high-intensity end. If you wanted to populate that space, you would have to adjust your exposure up. It doesn't matter whether the JPEG processor clipped the data. I don't know how you ran it. But the fact that it clipped the data, whereas the raw data aren't clipped, simply means that you need to go back and reprocess the raw data, so that the JPEG result doesn't show clipping.

OK -- just for the sake of inquiry I took a one of the photos (DR400) in which the EXRII processor did clip the highlights. I made sure there was no clipping in the raw file and I reprocessed that raw file in the camera where I set the push/pull factor negative to prevent the highlight clipping and of course equally darken the shadows (in other words underexpose the shadows to start and then lift them and then darken them again).

In effect then I'm using the camera raw processor to apply a negative exp. comp to undo the positive exp. comp. set when the photo was taken. I did see a slight reduction in shadow noise doing this.

You are not applying a negative exposure compensation when processing the raw file, and the camera didn't apply a positive exposure compensation when it took the photo. On the contrary. If you want to look at it this way, the camera applies a negative exposure compensation when using a DR expansion mode. It reduces the exposure compared to DR100. What you did in your raw processing is reduce the brightness of the image, but only compared to the JPEG file that you originally got out of the camera (with whatever JPEG-processing parameters you have specified. You haven't mentioned those yet).

I specified those in the first post with the photo.

Fair enough. To get the best representation of what the raw data will look like you should use Provia. But that's probably not a big deal here.

I don't care whether the term EC is used or not.

It is important. 'Exposure' refers to how many photons are collected. The processing, including the ever so important application of the tone curve and adjusting brightness, is what you do with the photons afterwards. And that seems to be the crux of the problem here.

What I did was take every exposure in 1/3 stop increments over a multi-stop bracket set -- I've said that. You told me I didn't set the exposure correctly. Well the exposure 1/3 stop more than the one I selected has highlights clipped by the EXRII processor. You're telling me that one was correct instead? Or even more exposure with more clipping was correct? You're telling me 1/3 stop less was correct instead? Which was it? You're telling me that if I had adjusted the JPEG parameters differently then the EXRII processor wouldn't have clipped the highlights in the 1/3 stop more exposure? I set all the processing parameters as far as they would go to avoid highlight clipping when creating a JPEG.

You're telling me that shooting a full bracket set in 1/3 stop increments over a 3 stop range will fail to produce a correct exposure even when, after examination, the most exposed of the set has severely clipped highlights and the least exposed doesn't have highlights. In doing that I still didn't manage a correct exposure according to you? I'm going to call you wrong on that.

I think we are going around in circles here. You showed the image where there was just no clipping in the ooc-JPEG version. Then you examined the corresponding raw data and complained about wasted levels. I told you that, if you wanted to utilize the higher levels, you need to adjust the exposure up. So, in that sense, for the image you presented, the exposure was not optimal/correct, if you want to maximize exposure.

The image I showed represented the maximum amount of exposure possible given the limitations of the EXRII processor. Assuming DR mode (that's what I was testing) and therefore assuming an OOC JPEG and not after-processing of a raw file, no additional exposure is possible without highlight clipping (JPEG parameters set accordingly). Noticing then the wasted sensor capacity, I say dumb kludge.

You can't tell me to increase exposure -- the EXRII processor will clip the highlights -- no way it won't. Of course I can capture a raw file and process it. I wasn't testing that.

Perhaps in your bracketed set, there is an image that does maximize the usage of the levels, and where the raw data aren't clipped but your in-camera processing caused clipping in the JPEG version. That would be the raw data to work with if you want to get the highest signal-to-noise.

Of course, it is quite normal for JPEG engines to lead to clipping despite the fact that the underlying raw data aren't clipped, because the in-camera raw processor just isn't that capable. Using an external processor will help. But if you are an ooc-JPEG shooter, then you will need to dial back on your exposure to not get clipping in the JPEG version. But then you'll have to live with the 'wasted' levels. Can't have both.

Got that. The EXRII processor isn't that capable and so when using the DR mode for OOC JPEGs it's required that we live with wasted levels to avoid clipping. Got that -- I think that was my point. That's why I called it a dumb kludge in the first place. That's why I'm still calling it a dumb kludge. You called it quick and easy -- I prefer a more colorfully descriptive term.

That original JPEG might not even be a good reference, depending on the processing parameters. For example, check what you get when you vary shadow and highlight controls. Perhaps that's already enough to avoid clipping in JPEG. But see below about dynamic range.

HOWEVER, since the DR function exists as a feature of the EXRII processor and is designed to create JPEGs, what I need you to do now is tell me how I can set the camera to apply a positive exp. comp. for the exposure while at the same time setting a negative exp. comp. for the EXRII processor. The whole point of the DR function is to get a SOOC JPEG and so it should be judged on how well it performs that function.

The DR expansion modes aren't necessarily designed only for JPEG. They are designed to avoid overexposure in general. And let's not use 'exposure compensation' here.

The DR expansion modes are specifically designed for the EXRII processor which exists only to create JPEGs.

There are two aspects here. The first one is what happens before you take a shot with a DR expansion mode, namely the camera adjusts exposure down and ISO up, which will of course be reflected in the raw data. As such, the DR expansion modes are entirely general and not specific to JPEG. The second one is what the in-camera raw processor then does with the data. A DR directive tells the processor to apply a special tone curve. If you manually set the exposure parameters and kept DR at 100, the tone curve would be different (e.g., shadows won't be boosted as much).

Fine, we can't say EC, let's use an external light meter or lick our finger and hold it up in the light or even better, in tricky situations like direct backlight, just do what I did and take a full bracket set in 1/3 stop increments. Then examine that bracket set. Da*mn! I must have missed the correct exposure between two of those 1/3 stops.

Perhaps. Check in RawDigger to see if you have it in your set.

What you need to do in the first place is proper metering. You do get a useful histogram before you take the shot. You should therefore get a pretty good idea already whether there will be clipping or not, under the active JPEG-processing parameters. Second, play with shadow and highlight controls, as already mentioned. Lastly, many scenes will simply exceed the dynamic range of a sensor. The X-Trans has a DR of about 9.5 EV.

So again I don't need to hear about proper metering. Many posts back I noted what I've again noted here: I shot a full bracket set and so did not fail to get a correct exposure.

The DR expansion modes reduce that DR (despite their name).

YES!! DUMB KLUDGE!!! Thank you.

As I earlier admitted I ran my test of the function without first investigating or thinking about how it works and what it does. I now understand. I am clear now about this -- no confusion here.

I did not fail to get a correct exposure, I got every exposure. The JPEG processing parameters aren't magical -- I used them to their full extent and in the manner appropriate when trying to photograph a scene with extreme lighting contrast. I picked a scene with extreme contrast as an appropriate test. I'm aware of dynamic range limitations, that's why I picked that scene -- it was an appropriate test. And I have now seen the limits of the DR function on my X-E2.

And given a scene with extreme lighting contrast where every bit of dynamic range from the sensor can count toward a best possible photo, starting the process by, as you noted, reducing that sensor DR is.........dumb.

Only if you shoot raw. If you like ooc-JPEGs, these expansion modes make sense, because the tone curves applied are specific to high-dynamic range scenarios.

I like best possible photos, but I'm not opposed to useful electronic automation if it's well designed and helpful -- therefore my curiosity led me to test the DR function. I started the test without full understanding; now I understand. I'm not an engineer and I don't know if they can design a better mousetrap or not, but I'm sticking with my assessment that hobbling a sensor's full dynamic range in order to capture high contrast lighting scenes is just a dumb idea. I won't use it in the future, but I also won't stop calling it a dumb kludge. You can keep calling it quick and easy.

Many high-contrast scenes simply exceed the DR that you have at that point.

Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow