[Ian Stuart Forsyth]

All good but I just thought I would add that ETTR was only ever relevant at base iso.

It was only really relevant to a narrow band of years for sensors that are produced mainly by Sony between the K10d to the K1 time line.
Out side of them this is more of the reality





If we look at the newer sensors from Sony starting from around 2016 to their present newer sensors


It look like this, now if the K3mrkIII uses a newer sensor then there is a benefit to raising the iso as one would when ETTR

[Ian Stuart Forsyth]

TrueDof and OptimumCS - George Douvas iPad/phone only

While I like to use this ap for DOF calc's I like to use it with some cavities.

I always like to decide based on my captured resolution and the details I am looking to display, often times these ap tell us one thing while in reality they maybe influencing us to make incorrect decisions in how to use our equipment.

If we follow the instructions from these ap's they don't tell us what we need to know. For example for a 24mp and if I want to display the finer details in the grass I may need to keep my lens open @ ƒ/8 to capture this detail for the amount of MP's of that camera.
In doing this I could be accepting less DOF to capture that grass detail.

Here is where it gets tricky, if i go back to the same scene but this time using a higher resolution camera for the very same amount of captured resolution grass detail but using this higher resolution camera at ƒ/16 I don't need to sacrifice DOF for that same captured resolution.



If we look at it @ ƒ/14 it tells us we have a blur spot of around 28 but in reality one camera can deliver more resolution stopped down even further than the camera with less MP's

[Rondec]

Your perception of ETTR matches mine.
And yes in high dynamic range situations you need to override it (and clip the highlights a bit) if you value the noise quality of the dark regions over the tonal range of the highlight regions.
But apart from that decision ETTR is still best practice in contrasty images isn't it?
I don't understand how that is relevant to the cardinal bird scenario.

I find the "histo is of the jpg" thing a bit overblown. But then I am always in Raw (all manual) and I leave my jpg variables alone and have highlight and shadow correction turned off. This means the relationship between the jpg histo and the raw remain fairly constant. Mostly the amount of "spare" in the histo at the clipping end is a nice margin of error if you are pedantic about highlight clipping (like I am!!)

I'm putting words in his mouth, but I think he is saying that he is struggling with shutter speed -- if he keeps it low the bird is blurry and if he lets it go up then his iso goes up (or he under exposes a bunch) and his camera doesn't handle the red hues very well at anything but base iso. At least that is what I am guessing he is saying.

Red is a special case and it is easy to blow the red channel, even when the histogram looks OK.

[TonyW]

While I like to use this ap for DOF calc's I like to use it with some cavities.

I always like to decide based on my captured resolution and the details I am looking to display, often times these ap tell us one thing while in reality they maybe influencing us to make incorrect decisions in how to use our equipment.

If we follow the instructions from these ap's they don't tell us what we need to know. For example for a 24mp and if I want to display the finer details in the grass I may need to keep my lens open @ ƒ/8 to capture this detail for the amount of MP's of that camera.
In doing this I could be accepting less DOF to capture that grass detail.

Here is where it gets tricky, if i go back to the same scene but this time using a higher resolution camera for the very same amount of captured resolution grass detail but using this higher resolution camera at ƒ/16 I don't need to sacrifice DOF for that same captured resolution.

If we look at it @ ƒ/14 it tells us we have a blur spot of around 28 but in reality one camera can deliver more resolution stopped down even further than the camera with less MP's

I agree we also need to apply some thought to output planning if we are to use these apps. in the field vs just winging it.

I actually prefer Lumariver Depth of Field by Anders Torger when I feel the need to confirm what I think I need to confirm adequate DoF. This app. IMHO offers some better options including saving different cameras and individual lenses using calculations based on actual camera sensor pixel pitch, It also allows to save cameras not in its database. Plus other options for calculations, CoC, Airy Disc, pixel pitch etc. I know that I am sounding like an advertisement for this but I do really like it

---------- Post added 03-13-21 at 04:12 AM ----------

It was only really relevant to a narrow band of years for sensors that are produced mainly by Sony between the K10d to the K1 time line.
Out side of them this is more of the reality





If we look at the newer sensors from Sony starting from around 2016 to their present newer sensors


It look like this, now if the K3mrkIII uses a newer sensor then there is a benefit to raising the iso as one would when ETTR

Good points about sensor changes! I would also add that ETTR at base ISO relating to noise was not always the case for every mnfctr. I remember seeing some Canon ? Images where the much higher ISO image had much lower noise than the base ISO image of the same subject shot under identical conditions

[swanlefitte]

At least that is what I am guessing he is saying.

Correct.
I have to find a way to make the world conform to my camera before I ever have the luxury of ETTR.
Cardinals in the bush make it frustratingly clear. I was just noting that. If it clips on both sides you can't. If it just fits but motion blur forces your hand you can't. A uniformly dark room is where ETTR works best. But in that case would we even consider that we were?

[biz-engineer]

It was only really relevant to a narrow band of years for sensors that are produced mainly by Sony between the K10d to the K1 time line.

I would rely on those chart when I see that the PDR curve of Pentax K1 mk II exceed that of a Phase One camera at high ISOs. Calculations can sometimes lead to unrealistic (and unreal) conclusions.

---------- Post added 13-03-21 at 19:05 ----------

Diffractionlimited to f/3.2

There is a lot of confusion around diffraction "limit". There is no such things as diffraction limit, because an image formation is impacted by various contributions, glass, aperture diffraction, digital sensor resolution, pixel level performance. Diffraction is one of the contributor, its amount varies mainly with lens aperture, but it never 0% nor 100%. Even at f1.0 is diffraction present and become more prominent with higher aperture numbers. Personally I prefer 100Mpixels with 12bits depth over 50Mpixels with 14bits depth, because displays do no more than 10bits depth per channel and the depth is limited to 8bits for printers. Printers aren't limited for resolutions however, as more Mpixels can be printed larger. However, if an image is 14bits deep per pixel/per channel, the extra bits over 8bits are lost when printed. So if I was given the choice between a 250Mpixels camera with 10bits per channel and a 25Mpixels camera with 16bits depth per channel, I'll take the 250Mp/10bits camera.

[TonyW]

.......
There is a lot of confusion around diffraction "limit". There is no such things as diffraction limit, because an image formation is impacted by various contributions, glass, aperture diffraction, digital sensor resolution, pixel level performance.

Oh yes there is
The diffraction limit is a function of the whole imaging medium e.g. smaller photosites or film with smaller grain will have a lower limit of diffraction than larger photosites or grain.
Diffraction happens at every aperture as light passes around the edges of the diaphragm. In optics the effect is known as the Airy disc pattern and describe the best focused spot of light that a perfect lens can make limited by the diffraction of light.

Diffraction limit is the point where the airy disc grows large enough to affect more than a single photosite. This is the meaning of Diffraction Limited

Diffraction is one of the contributor, its amount varies mainly with lens aperture, but it never 0% nor 100%. Even at f1.0 is diffraction present and become more prominent with higher aperture numbers.

Diffraction is present as you state at all apertures, but the point is at some stage depending on the pixel pitch of the sensor you will have reached the Diffraction Limit at a certain f/stop

Personally I prefer 100Mpixels with 12bits depth over 50Mpixels with 14bits depth, because displays do no more than 10bits depth

IF all you are doing is looking at a monitor or for that matter a TV then either is fine. Additionally your monitor will probably not be able to display 50 or 100 MP images without downsize interpolation. An 8K UHD monitor (7680 × 4320 pixels) is only 33MP! And while you are talking about bit depth current displays mostly limited to 10bit but 12-bit have been around for a while and one would expect a move to mainstream in the future

... and the depth is limited to 8bits for printers.

No it is not. For some time now Mac OS and printer drivers have offered the option of 16 bit printing. Latterly Windows systems have allowed 16 bit through XPS drivers

Printers aren't limited for resolutions however, as more Mpixels can be printed larger.

Printers are limited for resolution at fine settings of 720ppi or 600 ppi (and possibly 1440 and 1200 ppi). Indeed more MP can be printed larger or more MP can mean printing with a higher resolution than the standard 360 or 300 ppi taking advantage of detail in an image that may be present but not appreciated at a so called standard res.

However, if an image is 14bits deep per pixel/per channel, the extra bits over 8bits are lost when printed.

Not really for a couple of reasons:
If you edit your images (12,14 or 16 bit) in Prophoto for instance you should be in 16 bit (PS actually is 15+1 bit) your image will be sent through the printer profile and driver to the printer. At the editing stage it is quite likely that you have used the full 14 or 16 bits to refine your image and that edited image is quite happily sitting within the limits of 8 bit.
Should you really need more than the 8 bit image then you would select that at the printing stage through the print driver. A general consensus seems to be if your image has come from a 14 or 16 bit raw capture that there is very little to be gained most of the time printing in 16 bit. However if you are looking at computer generated images there may be some benefit in certain cases.

---

Last edited by TonyW; 03-13-2021 at 03:26 PM.

[Ian Stuart Forsyth]

I would rely on those chart when I see that the PDR curve of Pentax K1 mk II exceed that of a Phase One camera at high ISOs. Calculations can sometimes lead to unrealistic (and unreal) conclusions.

One of the problems is that the iso setting listed on the bottom is not base on the exposure size so indeed they can be deceiving in that one would assume they are based on the size of the exposure being equal while most of the time they are not. When we normalize shift the bottom scaling for the exposure size then the data is relatively accurate.

The only reason why I introduced the shadow improvement graph was to show that with many cameras as you increase the iso that in fact can reduce noise and also increase the DR that camera can produce when you are shutter speed limited.

---------- Post added 03-13-2021 at 11:55 PM ----------

Red is a special case and it is easy to blow the red channel, even when the histogram looks OK.

Most of the time for the lighting conditions we normally shoot in the red channel is the furthest from clipping if you are shooting raw.

Most of the clipping is done by the combination of the color space and how the raw data is being converted into a workable color space mainly what happens with the white balance, One only has to look at the white balance setting to see as to what is happening

For normal WB in daylight shooting conditions the Red channel is usually under exposed by about 1 stop difference to the green channel, so in fact the green channel is the one that can give us the most problems well before the red channel. What this means is that the green channel usually as no offset while the red channel can be offset at least 2 times to achieve the correct WB

Here is an image that was ETTR


Here is what the raw file looks like



Many times as to what people associate the reds in an image to clip are the furthest from clipping, most of the time where we see clipping starts in the whites followed by the yellows and then the reds

[eddie1960]

Interestingly I just saw the latest ACR update includes a super-resolution option that provides a post-process pixel shift-type idea. just a quick look at a 65:24 GFX 5oR file would let me print with 200ppi at 81" x 31" ... I guess I can stop looking for the xpan I crave (though TBH I am sure an Xpan neg drum scanned could exceed the file I got). I definitely will be shooting more in this aspect ratio as a starting point (still a 25mp file on its own without the super res option), what it does do is make upsizing with quality incredibly simple, and works with Raw Tiff Jpeg and outputs Raw Tiff jpeg and PSD. Every time I think I will dump Adobe for my Capture One 21 they find a hook to keep me.

https://www.dpreview.com/news/0299184261/adobe-releases-photoshop-for-m1-mac...olution-in-acr

[BarryE]

Interestingly I just saw the latest ACR update includes a super-resolution option that provides a post-process pixel shift-type idea. just a quick look at a 65:24 GFX 5oR file would let me print with 200ppi at 81" x 31" ... I guess I can stop looking for the xpan I crave (though TBH I am sure an Xpan neg drum scanned could exceed the file I got). I definitely will be shooting more in this aspect ratio as a starting point (still a 25mp file on its own without the super res option), what it does do is make upsizing with quality incredibly simple, and works with Raw Tiff Jpeg and outputs Raw Tiff jpeg and PSD. Every time I think I will dump Adobe for my Capture One 21 they find a hook to keep me.

Adobe releases Photoshop for M1 Macs and introduces Super Resolution in ACR: Digital Photography Review

Well it is truly a impressive feature, after initial testing.

I've enhanced some K-1 images up to 144MP and compared with PS bicubic it's much better. The RAW files show no obvious artifacts and they can be pushed just the same as the original. This is one of those features, if it is really as good as it first appears, that could change many aspects of photo processing and camera purchases, eg a heavily cropped 8MP image enhanced to 32MP or a K-1 to 14720x9824 for a massive print at 300ppi.

[Michael Piziak]

I do wonder if others also thought the new k3iii would have more megapixels (than 25/26)
Although it has been demonstrated, in this thread, that more aren't needed, but Pentax (and competitors) have been at 36 and higher...

[Serkevan]

I do wonder if others also thought the new k3iii would have more megapixels (than 25/26)
Although it has been demonstrated, in this thread, that more aren't needed, but Pentax (and competitors) have been at 36 and higher...

The only sensor over 26MP for APS-C is the 32MP Canon in the 90D and M6ii, so no chance of that happening.

[Michael Piziak]

The only sensor over 26MP for APS-C is the 32MP Canon in the 90D and M6ii, so no chance of that happening.

I didn't know that - thanks!

[bobbotron]

I find the 24mp on my K3 really good. I'm more excited that FF and MF sensors are becoming more affordable. I have a lot of different film camera including some medium and large format, and I'm way more interested in jumping to bigger sensors.


I can scan a darkroom 8x10 wet print and get an absolutely huge tiff, but... it's real diminishing returns.


Edit, so here's a trip.... what would get me really excited is something like a digital Pentax 6x7 with a 40MP CMOS sensor.

[texandrews]

It looks to me like something interesting is happening. At the one end, there's a move towards 100-150 mp. at the other, for several years now, there's been moderation at the aspc and video-centric cameras. Good reasons for both, but interesting to me.

We won't know until we see it, but if the K3III's IQ(including its DR and high iso) is what they suggest it may be, perhaps we've reached the spot where it's truly the quality of the overall IQ, and not the mp count that really matters. In fact, looks like we are there with apsc. If this camera turns out to be a stunner, then it will prove the position that apsc is the way to go for general photography for nearly everyone's actual needs (wants are a horse of a different color, as they say in the Emerald City).