[stevebrot]

Since it is from the jpg I am confident that if all I get is a few blinkies in pure white areas then using the RAW I am still safe. I have no idea how that would work on other cameras but it has been a good rule of thumb for me with the K-3II and K-1II.

Dear Pentax,
Christmas is past, but it would be nice to have a custom image option where the appearance and histogram reflect the capture data (i.e. gamma = 1, similar to linear TIFF) rather than a nice photo.

BTW...I have been a good boy except on Pentax Forums...


Steve

[Rondec]

That is correct as I understand it. And what I do. The important point is "up to the edge". If you have blown highlights you have blown highlights. Certainly underexposing is the safer choice in that regard. However, ETTR will in my experience give you more detail to work with and less noise. But if you go too far you blow the highlights irrecoverably. That is one of the reasons I use the histogram on the camera. Since it is from the jpg I am confident that if all I get is a few blinkies in pure white areas then using the RAW I am still safe. I have no idea how that would work on other cameras but it has been a good rule of thumb for me with the K-3II and K-1II.

Sure. I have just found that I have a lot more leeway on the shadow side of things than on the highlight side of things. I don't find that with, say, a sunrise image, I can push the exposure up at all and typically if I leave things at 0 EVs, I will lose some of the highlights. Probably if I used spot metering rather than the matrix metering, I would get a bit more accurate reading.

[TonyW]

Sure. I have just found that I have a lot more leeway on the shadow side of things than on the highlight side of things. I don't find that with, say, a sunrise image, I can push the exposure up at all and typically if I leave things at 0 EVs, I will lose some of the highlights. Probably if I used spot metering rather than the matrix metering, I would get a bit more accurate reading.

Typically in raw capture you can expect to be able to apply +3 - +3.5 EV from a metered ROI (spot reading, not Matrix or Centre Weighted) before highlight clipping, representing a camera metering calibration of around 12% - 12.5% leaving you with around -8 EV to give you a usable DR of 11 EV with a 14 bit Capture. These figures are approx and would need confirmation by testing your system.

If at all possible a spot meter reading on a highlight where you require detail is preferable adjusting the exposure to take into account the clipping point. For landscape work my preference is for a 1 degree hand held spotmeter as the camera spot reading is too wide and more difficult to judge the actual area covered.

[TonyW]

Dear Pentax,
Christmas is past, but it would be nice to have a custom image option where the appearance and histogram reflect the capture data (i.e. gamma = 1, similar to linear TIFF) rather than a nice photo.

BTW...I have been a good boy except on Pentax Forums...


Steve

I know/suspect your post somewhat tongue in cheek.

It seems to me that there is scant interest from the general photo public in wanting any form of raw data histogram for todays digital cameras (I think there are some exceptions - Leica?) even though it should be easy to accomodate.

3 Years ago Guillermo Lujik (known for promoting UniWB) started a petition via Change.org. Having spoken to him about software he had written to display raw data as histograms he pointed me to this petition.

Sadly in 3 years only 1,191 of us have supported this
Petition · Include RAW histograms on digital cameras · Change.org

[stevebrot]

3 Years ago Guillermo Lujik (known for promoting UniWB) started a petition via Change.org. Having spoken to him about software he had written to display raw data as histograms he pointed me to this petition.

I signed some time back. Guillermo's site is one of my bookmarked references.


Steve

[jatrax]

I don't find that with, say, a sunrise image, I can push the exposure up at all and typically if I leave things at 0 EVs, I will lose some of the highlights.

In that situation (Sunrise) my version of ETTR is actually going to be a negative compensation. I am exposing as far to the right as I can. Which might be 1 or 2 stops to the left. Makes total sense, right.

Though generally for sunrise I am going to bracket that and combine in post.

[stevebrot]

Though generally for sunrise I am going to bracket that and combine in post.

I would add "...or shot with a grad", but HDR in post is so very much easier and less expensive.


Steve

[MJKoski]

There is a real case for ETTL. If one uses Sony's compressed RAW, the highlight part of the file is heavily messed up. Pulling it back results in visible gradation of tonality in highlight areas.

ETTR is bad with Sony cRAW. Use uncompressed if optimal IQ and exposure is something you care about.

[TonyW]

There is a real case for ETTL. If one uses Sony's compressed RAW, the highlight part of the file is heavily messed up. Pulling it back results in visible gradation of tonality in highlight areas.

ETTR is bad with Sony cRAW. Use uncompressed if optimal IQ and exposure is something you care about.

Interesting, but I guess inevitable for a lossy compressed raw file. Not sure if I understand correctly what Sony does as I have never used one or even looked at file options, but my impression is that the camera captures data in 14 bit and then drops it down to 11 bit. The resulting loss can show posterisation which will if there be potentially magnified in post. Is that theory anywhere in the ball park?

[MJKoski]

Just open ETTR exposed cRAW with RAWDigger and check the histogram. It is has gaps in the hightlight-end. Now, one pulls back the exposure with the gaps and gets stairstepping gradients where there should be smooth tonal transitions.

No, it is not an issue in all files. Just those that should contain smooth skies for example. Or other large flat surfaces.

[gatorguy]

Right (!) or wrong I've always preferred to slightly underexpose and avoid blowing out any highlights at all for the most part. If needed shadow recovery for me is relatively easy and straightforward. I have far more issue when highlights lose detail.

[MJKoski]

Shadow pushing is nice...up to a certain degree. ETTR exposed RAW is better in shadow pushing too. Pentax allows massive shadow boosting with clean results when using in-camera multiexposures. 128x ME is good for massive push of 6-7 EV. Just watch out for shadow stepping which is the result of not having noise in the source material (14bits are not enough). Let the shutter click! Shutter likes to be used!

[TonyW]

Just open ETTR exposed cRAW with RAWDigger and check the histogram. It is has gaps in the hightlight-end. Now, one pulls back the exposure with the gaps and gets stairstepping gradients where there should be smooth tonal transitions.

No, it is not an issue in all files. Just those that should contain smooth skies for example. Or other large flat surfaces.

I do not doubt what you are saying at all. I do not have access to any cRAW images (unless that is the standard Sony output!). Just a quick look at a raw file from a Sony AR7 II. Seems to confirm a lot:

The Sony AR7 II file I look at in RD is certainly not what I would have expected to see from a 14 bit capture. Showing a Max value RGBG2 = 3546. A 14 bit capture you would expect to see values around 16384.

Seems to confirm that it is close enough to an 11 bit capture which would equal values of 2048 with the +7 bit that Sony state making up the rest.

The histogram shows it hitting the wall for the Green channels at around +2.5EV. The red appears about -1/2EV below that.

So applying ETTR as it was meant to be used at base ISO (and scene dynamic range that sat inside your cameras DR) if you metered an important highlight area you could apply a max exposure correction of +2EV - +2.5EV. Theoretically not clipping highlights.

But the problem remains in how the manufacturer decides to bin data and the most obvious place is in the highlight region as the first stop of exposure contains half of the data. So bin some of that and a user is left with the potential problem of editing and pushing the data too much causing misery