Stacking 8bit JPG input files to 16bit tiff as output

Images taken in a controlled environment or with a posed subject. All subject types.

Moderators: Pau, rjlittlefield, ChrisR, Chris S.

MiB
Posts: 68
Joined: Tue Dec 18, 2012 8:59 am
Location: Vienna
Contact:

Stacking 8bit JPG input files to 16bit tiff as output

Post by MiB »

Hello,
I tried to find an answer to my question in the older posts but did not find anything. Maybe it is a stupid question, but...

Would it make any sense if I would stack 8bit jpgs as input files to 16bit tiff as an output file?
Would such a workflow generate "more colors" in the output and be a benefit in postproduction compared to 8bit jpgs as input to 8bit tiff as output?

I work with tens of thousands of files, so shooting them in raw and convert them into 16bit tiffs not only takes many hours (shoutung in raw takes about twice the tiime compared to jpg and another 32 hours for the conversion from RAW to 16bit tiff in each of my current project(s)) but also creates terrabites of data for creating one output file.
The jpg-route to 16bit tiff output would save me lots of time and space on my HDD.

Thanks and best,
Michael
***
To collage tens of thousands of photos to create a more holistic view is my passion :)

rjlittlefield
Site Admin
Posts: 21214
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Would it make any sense if I would stack 8bit jpgs as input files to 16bit tiff as an output file?
Would such a workflow generate "more colors" in the output and be a benefit in postproduction compared to 8bit jpgs as input to 8bit tiff as output?
Yes, but you'll have to test to see how much improvement there is for your application.

The number of new colors that get generated can range from almost zero to quite a few, depending on stacking method and structure of the stack. With Zerene Stacker, and a typical stack, I would expect a lot more new colors from PMax than from DMap, and probably more colors with Brightness adjustment turned on versus leaving it off.

One view of the improvement can be seen by looking at the histograms of 8- and 16-bit TIFF saves of the same PMax output image, after application of an extreme levels adjustment.

Image Image


Image

Image

Clearly the underlying PMax image has a lot finer gradation than can be captured in an 8-bit output.

However, despite the impression that we get from the histogram, visually there is hardly any difference between the two levels-adjusted images.

This is because there's enough pixel noise in the images that the colors are effectively dithered anyway. If I zoom in to 1600% and watch individual pixels as I flash between 8-bit to 16-bit source, then I can clearly see individual pixels getting brighter or darker depending on how they quantize. But if I zoom out to "only" 200% view so that I see a lot of pixels at the same time, then I have the impression that flashing between 8-bit and 16-bit makes no difference.

Bottom line, I cannot tell you what the answer will be for your application. I can only suggest to try it yourself and see what you find. Please let us know too.

--Rik

MiB
Posts: 68
Joined: Tue Dec 18, 2012 8:59 am
Location: Vienna
Contact:

Post by MiB »

Thanks, Rik, for your prompt reply and the examples!

I did some other experiments on 8bit jpgs as source files and 16 bit tiffs as source files, stacked into 16 bit tiffs.

The stunning results are (for me at least), that there is almost no visible difference whether I use 8bit source files compared to 16 bit source files. Perhaps this has to do with my 8bit per channel display on my notebook. I will check on a native 10bit monitor soon and might also do test prints, to see if one would see any differences in a printed photo.

Rendering out 16 bit tiffs from the stacking process does not make a big difference for me because it is just around 7GB each project compared to around 3TB when working with raw files converted to 16 bit tiffs as source files instead of shouting jpgs.

A 1200% crop with two differnt source bit-rates:
Image
Which had the 8 bit jpg source and which the 16 bit tiff source?

I uploaded a larger area multi-layer tiff where each layer has different source file formats.->
https://1drv.ms/u/s!Am4tkQLuFgrEzFXz8Qt ... V?e=TxvN8N
(please copy the whole line into your browser; the cropped file is a little lager than 1GB the other one more than 2GB, but I is really interesting to swap the layers at different magnifications). If someone has problems downloading the file)s) from onedrive please let me know then I would try to upload it to wetransfer.
There are some smaller parts whre one would see a difference.

The upper layers are already post processed ones (sharpness, contrast,…). On my notebook at 100% there is no difference. At 400% one could see that there is a little more info in the 16bit sourced ones. Also, the exported jpgs are just over 10% bigger when 16bit source files have been use for the stacking. So, there is more data in the files, but I just cannot see it (at least on this display).

In my opinion for my subjects and backgrounds it makes absolutely no sense to work with 16bit source files or raw shooting.
***
To collage tens of thousands of photos to create a more holistic view is my passion :)

Guppy
Posts: 148
Joined: Mon Dec 20, 2010 10:36 am
Location: Switzerland
Contact:

Post by Guppy »

Hi, Michael.

Focus Stacking images often require a lot of post-processing, dark areas are brightened.
8 bit images then show less structure than 16 bit images.
16Bit images have more tonal different levels within one f-stop than 8Bit images.
If the brightness of 8 bit images is changed, tonal value breaks down faster than with 16 bit images.

Simple test:
A picture taken in 12Bit or 14Bit RAW format is developed into the 16Bit TIF format.
Then this picture is darkened by 9 f-stops and saved, it is black.
then this picture is brightened by 9 f-stops (result 16 bit).

For comparison, the original is converted to 8 bit and then also darkened by 9 f-stops and brightened again by 9 f-stops (result 8 bit).

Compare result 16 Bit with result 8 Bit.

The 16 bit image leaves more room for editing.
---------

Focus Stacking Bilder müssen oft stark nachbearbeitet werden, dunkle Stellen werden aufgehellt.
8 Bit Bilder zeigen dann weniger Struktur wie 16Bit Bilder.
16Bit Bilder besitzen innerhalb einer Blende mehr Tonwertstufen wie 8Bit Bilder.
Verändert man die Helligkeit bei 8 Bit Bildern können schneller Tonwertabrisse entstehen wie bei 16Bit Bildern.

Einfacher Test:
Ein Bild in 12Bit oder 14Bit RAW Format aufgenommen wird in das 16Bit TIF Format entwickelt.
Dann wird dieses Bild um 9 Blenden abgedunkelt und abgespeichert, es ist schwarz.
dann wird dieses Bild um 9 Blenden aufgehellt (Ergebnis 16 Bit).

Zum Vergleich wird das Original in 8 Bit umgewandelt und dann ebenfalls um 9 Blenden abgedunkelt und wieder um 9 Blenden aufgehellt (Ergebnis 8 Bit).

Vergleiche Ergebnis 16 Bit mit Ergebnis 8 Bit.

Das 16 Bit Bild lässt für die Bearbeitung mehr Spielraum.

Kurt

MiB
Posts: 68
Joined: Tue Dec 18, 2012 8:59 am
Location: Vienna
Contact:

Post by MiB »

Hi Kurt,

thanks for your message!

Yes, I understand. That´s why I also attached some posprocessed images in the layered tif. I will have to do other experiments whith the 16bit output from the stacking process - especially with brighness modifications. As Rik stated, the stacking process should produce more colors and working with raw for my projects is just not really practicable.

Besides I´m a big fan of your work. Your scale photos are phantastic.

Cheers,m.
***
To collage tens of thousands of photos to create a more holistic view is my passion :)

Post Reply Previous topicNext topic