"Meters Don't See 18% Gray"

[PopsPhotos]

Colin, that is why I posted only the pertinent portion of Ken's thread. I have heard that some cameras and some photographers don't have the look they want when they use the Expodisk. I haven't tried it, so I can't comment on it directly.

Pops

[joelk]

Agreed.



I think Guillermo posted an excellent article (with example) on this here ...

Is the end of bracketing for HDR nearing?

That's neat. This guy is exposing to the right to maximize his dynamic range - check it out:

He stopped the camera down to 80 ISO. That means you're going to be giving the camera 2 stops more light than you would have at ISO 320.

Forcing a low ISO is exposing to the right by definition. He can say he underexposed by 6 stops but he added two stops of light by rating at 80. If he rates at 320 he can only underexpose 4 stops. That's why he says higher ISO's behave typically. That camera may have a sensor that truly should be rated at 80 in order to maximize its' signal. This guy's test basically proves that.

SO - ISO 80 is maximizing this guy's signal to noise ratio - which is exactly what you need to do to squeeze all 12 stops out of a camera. Basically you nail the exposure without underexposing at all and THEN you get 12 stops. If you underexpose by two stops then you've only got 10 stops.

Underexposing to be safe is tossing away information you could have you used for the single shot HDR (also known as tone mapping). Again - it goes back to what John Knoll said. If you want to get 15 stops out of a camera it's only going to happen with a an optimal exposure. And that puts you in danger of clipping highlights.

Guys will probably buy a 12 stop camera and routinely only get 9-10 stops out of it because they underexpose to be safe. Probably a good plan.

Tone mapping in camera is coming.

The guys at RED.com have put a version called HDRx into their new movie camera called EPIC and are claiming 18 stops of range. They are actually grabbing two exposures at once and putting them together on the fly to extend a 12-13 stop chip. That can introduce some artifacts because it's a moving picture camera.

The Hobbit is being shot on that camera in 3D. Brian Singer wants to shoot his next movie on that camera using much more natural light than is typical in movies.

re: The expo disc... haven't used it but I remember not liking the concept because I want to know black, white, gray and some saturation reference at the subject's position. That's a color chart. I shoot raw so I can white balance in post off of the chart. Nailing it perfectly in camera is tough. Ansel Adams used a darkroom for a lot of his magic.

That said, every studio shot at the same exposure in the same light at the same distance with the same lens should all look like each other.

[Colin Southern]

That's neat. This guy is exposing to the right to maximize his dynamic range - check it out:

I'm confused - if it's 6 stops under-exposed, it's not ETTR. Or are we talking about something else?

He stopped the camera down to 80 ISO. That means you're going to be giving the camera 2 stops more light than you would have at ISO 320.

I'd say it's pretty much a "given" that if you want max DR then you shoot at lowest possible true ISO.

If you underexpose by two stops then you've only got 10 stops.

Which is still about 4 more than the average reflective scene needs.

Underexposing to be safe is tossing away information you could have you used for the single shot HDR (also known as tone mapping).

My opinion is that "single shot HDR" is an oxymoron.

Again - it goes back to what John Knoll said. If you want to get 15 stops out of a camera it's only going to happen with a an optimal exposure. And that puts you in danger of clipping highlights.

yes, but then again, no real surprises there.

The Hobbit is being shot on that camera in 3D. Brian Singer wants to shoot his next movie on that camera using much more natural light than is typical in movies.

Yeah - that's over my way (we've just had a big scrap over actors contracts etc). I hate to say it, but The Hobbit has all the appeal of LOTR, which has all the appeal of a root canal at the dentist for me I'm afraid. Actually, that was a bit exaggerated ... at least at the dentist they give you something for the pain ...

Avatar was good though

[JonathanC]

How bizzare.
6 stops underexposed = very dark image.
PP levels adjustment = lightened image. (with lots of noise)
PP median filter and downsize to suit web = acceptable image.
The point being?

Let's assume I have the wrong end of the stick here, and that the guy is doing the 6 stops under bit to show that, or test if, his camera can do low noise shadow detail. This would be a very valid way to do this and this alone.

If the exposure could be closer to "correct", then the shadow detail would have less noise just because there would be a lot more signal and a bit more noise.

I, and lots of others, have been doing this for years when doing astrophotography. The subject is very dim in the dark sky, the f/ number is typically f/5 for a fast system and f/20 is not uncommon. The image stretching and median filtering the black bit is the way it works. Nothing to do with HDR, just recovering an underexposed image.

This possibly lost something between the guy writing it and me reading it, and I'm probably not getting the point. I thought I would add something for the purposes of discussion so that I, and perhaps others, can get a better grip on what was being described here.

[joelk]

Nothing to do with HDR, just recovering an underexposed image.

As to this guys 6 stop under thing, I think it's a weird test. A better test is to bracket a bunch of exposures, make them all look as good as possible in photoshop and post them all. Choose which one looks best. You can figure your optimal ISO from there. (Lowering ISO stops working at some point - I wonder if ISO 80 actually clips whites in the real world on that camera)

From there you can see how many shots below the optimal one are still looking very good in the shadows. Have some nice navy blue, burgundy etc. in the shot. Pulling a white shirt from 4 stops under isn't helpful.

"underexposed" is ultimatley in the eye of the beholder.

I think it's fair to say you can probably shoot these cameras 2 stops under optimal exposure and still get a nice image. I think that's been Colin's point and I agree.

Despite how much range a camera has some extra picky guys (like me) are still going to try to get the exposure as close to optimal as is reasonable to hold shadow detail. I worry noise reduction techniques may fool people into thinking they have more range than they really do. Just because there's no noise doesn't mean you didn't lose color information. I rarely see people talk about that.

[JonathanC]

"underexposed" is ultimatley in the eye of the beholder.

No it isn't, technically speaking. It's where the full dynamic range of the sensor isn't utilised i.e. no wells more than 80% full. That is, technically speaking.
In the real world you can shoot a black cat on a black rug, as previously suggested, but pulling that out of a 6 stops underexposed capture isn't going to be a lot of fun.
The sensor data should represent the image as far as possible, in that the bright bits come out as bright after all the processing. You can "underexpose" by a stop or two with the aforementioned cat, but that's only compared to a TTL meter reading. If you did an incident light reading, or a flash incident light reading and metered from that, then the cat would be as black as a black cat in the completed image. The exposure though, would be correct, even though the histogram was all piled up on the left.
Hope that made sense?

[joelk]

No it isn't, technically speaking. It's where the full dynamic range of the sensor isn't utilised i.e. no wells more than 80% full. That is, technically speaking.

Certainly well intentioned people disagree but I agree with you. I just use the phrase "optimal exposure" for the point at which the cup is full. Complicated lighting scenarios really depend on the photographers intent. Underexposed to one is a perfect silhouette for another.

Your cat example is perfectly explained. I'm assuming a correct exposure and not a fooled meter reading. Going back to my idea of having a white reference and putting that in Zone IX - the cat will automatically drop where it belongs down in Zone II or III and the exposure would be very good.

[Colin Southern]

How bizzare.
6 stops underexposed = very dark image.
PP levels adjustment = lightened image. (with lots of noise)
PP median filter and downsize to suit web = acceptable image.
The point being?

Let's assume I have the wrong end of the stick here, and that the guy is doing the 6 stops under bit to show that, or test if, his camera can do low noise shadow detail. This would be a very valid way to do this and this alone.

If the exposure could be closer to "correct", then the shadow detail would have less noise just because there would be a lot more signal and a bit more noise.

I, and lots of others, have been doing this for years when doing astrophotography. The subject is very dim in the dark sky, the f/ number is typically f/5 for a fast system and f/20 is not uncommon. The image stretching and median filtering the black bit is the way it works. Nothing to do with HDR, just recovering an underexposed image.

This possibly lost something between the guy writing it and me reading it, and I'm probably not getting the point. I thought I would add something for the purposes of discussion so that I, and perhaps others, can get a better grip on what was being described here.

I think perhaps you have missed Guillermo's point; he's just demonstrating the dynamic range of his new camera. In essence he's thrown away the best 6 stops of a potential capture, and STILL had at least 4 good ones left (and possibly a couple more) therefore it's a valid assumption that the camera had 10 (or perhaps up to 12) to start with ... and if you have perhaps up to 12 stops of DR available to start with then he's suggesting that this is possibly approaching the point where what may have traditionally required 2 exposures to capture the dynamic range of a scene, now the same DR may be able to be captured in 1 shot.

[Colin Southern]

You can "underexpose" by a stop or two with the aforementioned cat, but that's only compared to a TTL meter reading. If you did an incident light reading, or a flash incident light reading and metered from that, then the cat would be as black as a black cat in the completed image. The exposure though, would be correct, even though the histogram was all piled up on the left.

You'd be amazed at how many people DON'T get that. Shooting a black cat on a black rug with ETTR is not a pretty sight!

[Gingerbaker]

Just wanted to pipe in with the idea that if we are planning to shoot RAW, then we should go into our camera's setup and tone down the jpg contrast, brightness, color saturation settings to a neutral value or lower in order to give us a more accurate histogram readout as far as the RAW image is concerned.

I have done this and feel pretty confident that when I ETTR enough to get some highlight clipping on my histogram/LCD clip warning I can easily recover this information in ACR if it actually shows up as a problem. I shoot for some clipping warning, in other words, and then feel good to go.

[Colin Southern]

Just wanted to pipe in with the idea that if we are planning to shoot RAW, then we should go into our camera's setup and tone down the jpg contrast, brightness, color saturation settings to a neutral value or lower in order to give us a more accurate histogram readout as far as the RAW image is concerned.

The problem is though that some (well me anyway) use the in-camera JPEG to judge other things apart from the histogram (eg for studio shoots the camera spits out a low res jpeg which we display on a 40" LED TV, and we need to make lighting decisions based on what it could look like with the RAW we have - and that in turn means making adjustments to the jpeg contrast, brightness, saturation etc).

I honestly think that people worry about ETTR too much; sure, with a high dynamic range scene there's no alternative, but for normal reflective scenes what we need is so far above the noise floor it doesn't make a lot of difference if we throw away a stop or even two.