A rather stunning ETTR processing vid; pulls a picture out of basically a white sheet

[Inkanyezi]

In most devices, the highest level of linearity / accuracy tends to be in the middle portion of the device’s range and accuracy tends to drop at the extremities. So biasing the results so that they are weighted to the right of the histogram’s centre line would make sense, but how far is too far?

The quirk here is to determine what's the "middle" of the device's range. As we have an arithmetically linear device, but expect an answer that is geometrically linear, there are two different mathematical entities involved, and the "middle" in one of them is the division between the top two f-stops of dynamic range. Exactly half of the available data resides in the rightmost f-stop of the dynamic range of the device. Within this upper range of one single f-stop some less linear response might be expected in the analog device, the sensor, whereas the A/D converter is purely linear. So the actual "middle" of the range of both devices, both the analog sensor and its A/D converter is at the far right of the histogram. If nonlinear response is expected in the very last portion of this range, suffice to leave a minuscule gap at the right side of the histogram to bring the whole image into the linear range.

Pragmatically though, we know that at base ISO, we won't be disturbed by noise even in the middle range after conversion. So even if in theory it might be preferable to expose to the right, in practice, particularly at base ISO or low ISO levels, there is no need to cling strictly to the ETTR rule, unless there are very dark areas that we wish to lift into the visible range in the final image, as when using "D-Lighting" or similar technique.

At high ISO settings however, it becomes more important to expose to the right in order to avoid noise in the middle tones.

[Manfred M]

Urban - when I was thinking the middle portion of the range, I was thinking more of the area of the response curve with the highest level of linearity, as opposed to a mathematical middle of the curve. The analogue amplifier section built into each photodetector site is quite basic, because of size constraints, so I would expect to see the classic non-linear behaviour.

Once we get to the digital parts of the circuit, the behavior is much more predictable.

[Kris V]

Just for the fun of it: I tried it on a picture that I already trashed, and it really surprised me how much detail I recovered:
Here is the OOC - uncroppen and unprocessed:



and after:

[Manfred M]

Paul - Yesterday we were discussing the Aptina sensors and today I ran across this:

http://www.dpreview.com/news/2012/10...le-10MP-sensor

[ajohnw]

Below is a indicative illustration of zones compared to the reasonably linear response of a silicon sensor with a 12bit AD converted output. All the zone charts I have seen show equal length steps with the tone values doubling with each step. The illustration varies the length to reflect the space tone values may occupy related to the possible data conversion.




From a numeric point of view there seems to be tremendous room for detail to be held in zones 9 and 10 so it appears that ETTR is a valid approach. Theoretically zones 9 and 10 occupy even more room than is shown but it conveys the concept of the highlight headroom providing clipping does not start to happen.

Warning – the above assumes the AD conversion is linear and stored as RAW values but to speed up the conversion manufactures may not bother to completely resolve the higher values or may use a reference curve during the AD conversion. Why use so much time and storage for highlights as it is basically redundant precision? Examining the RAW file may give a clue.

I am curious as to how the designers approach it, so if any of you can provide factual insights I would be pleased to know.

In some ways that headroom is misleading. We can process it to make it visible to the eye as has been done in the video that started this post but we do not see in that fashion. We see light level changes logarithmically. Rather like the low end in the jpg you have posted. There is a bit about it on cambridgecolour https://www.cambridgeincolour.com/tu...-human-eye.htm. Opinions vary but basically we see equal light changes in an odd way sort of logarithmically but really following a power law. Best proof of that is the magnitude of stars well explained on this page http://www.skyandtelescope.com/howto...de_System.html . Most of the info about on that subject skips the basics. It shows percieved magnitude against actual brightness and explains it as well. We see like that where as a camera sensor tends to see linearly. A perfect sensor would see perfectly linearly as it counts photons.

Going back to the shoot to the right demo clearly the procedure can't work if there is any clipping and it could have been shot at a different exposure level. The photographer was probably capable of seeing up to 14 stops of light level if needed when viewing the scene. The total range in that respect is reckoned to be 20 or more with dark adaptation. The camera less. Think this through and the starting shot to the final result doesn't make any sense at all. Stinking fish in fact. It might only work because many modern cameras compress the high brightness end as well as the unavoidable compression at the dark end but that's for jpg use and is why they can have detail bought out. The other way it could work is that there isn't much light variation in the scene. It's then been shot to the right without clipping, tonal range is very low and displaying it all near the white level on a PC screen unsurprisingly doesn't show much at all. The contrast is then stretched from a couple of stops to near full house so it can't really look anything like the real scene. It's really an inverse of hdr. Vague reflections have become features etc. Fine if that is what some one wants to do. Always shoot to the right - most shots that advocate this contain whites and darker areas. It can help capture details in the darker area and highlights will be retained if they aren't clipped and as in this case contrast can be stretched to make something of what would probably be a very boring picture without it. This one though could have been stretched from the shoot in the middle or left I suspect any old place really. The important aspect is not to clip the highlights and maybe judging exposure from the histogram is better than using the metering. :-) Probably is as well.

-

[ajohnw]

Paul - Yesterday we were discussing the Aptina sensors and today I ran across this:

http://www.dpreview.com/news/2012/10...le-10MP-sensor

66db dynamic range doesn't sound too good to me. It's similar to 10bit cmos microscope cameras with 8bit output. Most compacts work on that format. 10bit a/d and 8 bit out.

Dpreview is a strange place. Reading the comments I really do wonder about Sony using Aptina sensors They have been making their own for rather a long time and abandoned further ccd development some time ago to concentrate on cmos sensors. One interesting thing about Aptina is that when it was Micron I could download full data sheets. Just told them I was interested in building a 1 off microscope camera.

-

[Glenn NK]

That was basically what I was thinking of when asking about your ISO setting. Note what they say there also: for most cameras
the base ISO is 100, Canon being an exception with base at 160 (as seen in the graphs you referenced).

As a side note, I find the opening image a touch misleading, in that the base gray gets a lot lighter with higher ISO values (in the video itself
the background is more constant, and the noise less visible!).

Remco:

Both my bodies are Canon - multiples of 160 are the order of the day so to speak.

Glenn

[Manfred M]

66db dynamic range doesn't sound too good to me. It's similar to 10bit cmos microscope cameras with 8bit output. Most compacts work on that format. 10bit a/d and 8 bit out.

Dpreview is a strange place. Reading the comments I really do wonder about Sony using Aptina sensors They have been making their own for rather a long time and abandoned further ccd development some time ago to concentrate on cmos sensors. One interesting thing about Aptina is that when it was Micron I could download full data sheets. Just told them I was interested in building a 1 off microscope camera.

-

Actually with Sony using Aptina, it really isn't necessarily all that surprising. Many industries, especially the tech industries you have many examples of companies who are normally competitors getting together in a customer /client relationship. Look at the Apple / Samsung relationship; the two are fighting each other in courts around the world, when it comes to phones and tablets, yet Samsung is Apple's largest single component supplier. Sony builts some of Nkon's DSLR sensors, even though both companies are in the camera business.

It may be as simple as a resource / capacity issue at Sony. We have no idea as to the number of wafers that their fab can turn out in a month, and we do know that Nikon is a very large user of the Sony sensor facility. Buying and off-the shelf commodity sensor may be the right thing for Sony to be doing for that part of their camera business.

[ajohnw]

Actually with Sony using Aptina, it really isn't necessarily all that surprising. Many industries, especially the tech industries you have many examples of companies who are normally competitors getting together in a customer /client relationship. Look at the Apple / Samsung relationship; the two are fighting each other in courts around the world, when it comes to phones and tablets, yet Samsung is Apple's largest single component supplier. Sony builts some of Nkon's DSLR sensors, even though both companies are in the camera business.
It may be as simple as a resource / capacity issue at Sony. We have no idea as to the number of wafers that their fab can turn out in a month, and we do know that Nikon is a very large user of the Sony sensor facility. Buying and off-the shelf commodity sensor may be the right thing for Sony to be doing for that part of their camera business.

You will see what is actually going on if you read the top of this page http://www.sony.co.uk/hub/twilight-football/5 - note the word selected.

Sony have always been fairly good at providing some technical info about their own parts see
http://www.sony.net/SonyInfo/technol...xmor_r_01.html

This illustrates a problem with the web. Some one reads sony using aptina sensors so they suddenly are using them in all cases. Personally I suspect it's a case of what they have always done for a long time - used their own sensors in the better cameras. 69db sounds like half decent compact type performance to me based on microscope cameras using Aptina sensors with a dynamic range of 61db. No idea how they work these out. Given noise levels in the microscope cameras it just doesn't make sense.

I bought a sony dsc-tx5 a short while ago 10.2mp exmor to replace my still working dsc-t5 5.1mp ccd one. Even though the ccd in that was one the best available at that time it's like comparing chalk and cheese.

One thing for sure though it's difficult to know who actually makes what in this field but it's usually clear who owns the technology used. Canon are reckoned to have developed and fabbed their own sensors from scratch. I have to wonder about that. Canon for instance have used Motorola parts in their cameras and still do as far as I am aware. :-) They were over the moon when they got that business. I know that from talking to the reps. That's microcontrollers. Who makes the sensors? Pass. Hitachi might have obtained the Canon business but they used to make Motorola parts under license. Motorola got upset because they could produce better parts than they could so Hitachi designed there own processor. Motorola kept taking them to court for stealing their technology so no one would dare to use the part in main stream products. The high volume electronics industry is an odd place and it's difficult to know who makes what or where it comes from.

-

[pnodrog]

Hitachi might have obtained the Canon business but they used to make Motorola parts under license. Motorola got upset because they could produce better parts than they could so Hitachi designed there own processor. Motorola kept taking them to court for stealing their technology so no one would dare to use the part in main stream products. The high volume electronics industry is an odd place and it's difficult to know who makes what or where it comes from.

-

Get over it John (just kidding) - that was over 20 years ago when Hitachi had to bring out the H8 and H16 because Motorola got the huff over Hitachi 6300 range competing with their 6800 range. I was involved in having to redesigning a range of products because of the supply problems. They both missed out - we changed to a Texas Instruments processor. Thanks for the memories.

[revi]

Remco:

Both my bodies are Canon - multiples of 160 are the order of the day so to speak.

Glenn

That's what I understood, but there are some misguided souls here that use other brands, so they should perhaps not automatically
pick 160, 320, etc. as their favourite ISo setting

Remco

(PS. I'm using Sony...)

[ajohnw]

Get over it John (just kidding) - that was over 20 years ago when Hitachi had to bring out the H8 and H16 because Motorola got the huff over Hitachi 6300 range competing with their 6800 range. I was involved in having to redesigning a range of products because of the supply problems. They both missed out - we changed to a Texas Instruments processor. Thanks for the memories.

Hitachi made a few parts identical to Motorola's under license. Basically there fab process was better so more and more people started to buy the Hitachi parts. We used 2 in a product and had to buy loads of them when the trouble started. We got fed up of Motorola interesting us in processors, doing initial work and then finding that Bat had changed the design to something GM wanted. Left me with the job of picking something for a brand new product far more complex than the previous ones with about 2 years to develop it and no chance of switching processors again. Got me a pay rise and we went to NEC.

-

[pnodrog]

In the garage I think I still have a tube of about 20 Hitachi 63701 OTP for emergencies. Just send a cheque.
Small world - similar problems but don't start me reminiscing I am hard to stop.

[ajohnw]

In the garage I think I still have a tube of about 20 Hitachi 63701 OTP for emergencies. Just send a cheque.
Small world - similar problems but don't start me reminiscing I am hard to stop.

It is hard to stop and only mentioned as a poor example of what goes on in the electronics world. And how effective injunctions can be even though there is probably no real basis to them. The core was nothing like anything Bat made.

-

[Shadowman]

Just to clarify a little, since my OP was so brief: I found this while searching for Lightroom tutorials. I think LR is a wonderful tool and I am certain I still only use a fraction of what it is capable of doing.
I was not looking for ETTR or ETTL information at all, and I am certainly not arguing in favor of ETTR (nor against it, for that matter).
I stumbled upon another short vid which happened to be ETT-LEFT in the extreme, the main point of which was to show viewers why they should use RAW. If I ever find that again, I may post it for its PP information (and for pointing out the value of shooting RAW).

The one thing I noticed from the link was the ease of use with Lightroom. I am sure the same recovery could be achieved with Elements or Photoshop.

[Colin Southern]

The one thing I noticed from the link was the ease of use with Lightroom. I am sure the same recovery could be achieved with Elements or Photoshop.

Just keeping in mind John that LR and the ACR portion of PS (ie the bit that one would be using to make the recovery) use the same engine and thus in essence have exactly the same controls ... so really just as easy to do it in ACR as it is in LR.

[Shadowman]

Just keeping in mind John that LR and the ACR portion of PS (ie the bit that one would be using to make the recovery) use the same engine and thus in essence have exactly the same controls ... so really just as easy to do it in ACR as it is in LR.

Colin,

That's what I suspected. I currently use PSE8 and tried the ETTR conversion and got similar results. LR just makes it look so much easier.

Thanks

Thanks

[Colin Southern]

Colin,

That's what I suspected. I currently use PSE8 and tried the ETTR conversion and got similar results. LR just makes it look so much easier.

Thanks

Thanks

I know a lot of people like LR, but it just seemed clunky and inefficient to me -- I prefer the stripped-down ACR interface. Bit like a BMW -v- a Ferrari; both are performance cars, but one is built for "luxury & performance" whereas the other is just built to "go like stink"!