Dynamic Pixel Shift - a simple comparison


LennyBloke

Link Posted 11/01/2021 - 16:35
A very quick and dirty comparison of the K1 MkII implementation of Dynamic Pixel Shift against normal shooting - both images are handheld and shot with exactly the same settings (1/50 @ f4.0 ISO3200) and the post processing in LightRoom uses exactly the same values (Highlights/Shadows etc.).

The focus was manual and focus confirmation was given aimed at the white circle with the small dots on the red wrapped chocolate. As you can see the composition isn't identical - this was due to me changing the menu settings and waiting for the in camera processing. Please click to see a bigger image with more detail.

1. With DPS



2. Without DPS



Probably not a great choice of subject but the light was fading so window-lit Xmas choccy leftovers it is. I'll do some more in more suitable condition when I can - but does this example convince you?


LennyBloke

Mike-P

Link Posted 11/01/2021 - 17:53
LennyBloke wrote:
but does this example convince you?

I brought them both up on my laptop side to side and then full page swapping between the two and the short answer is no.

Is there much difference in file size?

Edit : I tried one the other day using the K-1 and 50-135mm and was blown away by the quality, rendering and sharpness. Then I did it again normally and decided it was just that Pentax files are that good anyway as there was no difference to my eyes. However my eyes are not what they used to be
No equipment list here but thanks for taking an interest. My Flickr
Last Edited by Mike-P on 11/01/2021 - 17:56

Flan

Link Posted 11/01/2021 - 19:27
Hi LennyBloke and and Mike_P my tuppence worth , there is definitely a difference which is evident in the Gold sweet both left and right of the centre red sweet. they both contain more detail (to my ageing eyes) i put on my best focals and studied both images . also took my first pixel images today but they might not be good example as my lens was wet with drizzle and i underexposed the shot and i had a polarising filter on and it was a dank day . i will post in the gallery . LennyBloke thank you for starting the thread regarding pixel shift images. i hope others will have some input and make it a reference point thread. oh and MikeP the file size i got today was around 120 MB
Last Edited by Flan on 11/01/2021 - 19:43

Mike-P

Link Posted 11/01/2021 - 19:55
Flan wrote:
Hi LennyBloke and and Mike_P my tuppence worth , there is definitely a difference which is evident in the Gold sweet both left and right of the centre red sweet. they both contain more detail (to my ageing eyes) i put on my best focals and studied both images .

If you look at both shots carefully you will see they were taken at slightly different angles so the writing on the sweet to the right of the red in the non pixel shift picture is slightly covered by the rim of the glass bowl. I am presuming they were taken hand held (as per the other thread) and this sort of test needs a tripod to rule out movement of any kind.

As for the file size, I was wondering about the final .jpeg
No equipment list here but thanks for taking an interest. My Flickr

LennyBloke

Link Posted 11/01/2021 - 20:27
Mike - your question on file size - the one with Dynamic Pixel Shift is 18,214Kb and without is 16,663KB, I've just uploaded the full size images to Flickr in case you want to pixel peep
With: https://www.flickr.com/photos/62460860@N05/50825942161/in/dateposted/
Without: https://www.flickr.com/photos/62460860@N05/50825941391/in/photostream/

And you're right about the composition of course and yes they were hand held but this test cannot be done with a tripod as this is the Pixel shift mode that is designed for hand holding - if my understanding is correct it relies on the photographers normal movement between each frame to do its pixel shifting (the original Pixel shift moves the sensor by one pixel between frames - that mode is the one for use on a tripod).

Flan, I think my test isn't good enough to be conclusive - the slight difference in position and of course the focus could have changed very slightly (focus confirmation does seem to "indicate" over a small range rather than just one precise point).
It'd be great to see your images - you've taken a different approach to me in using RAW, and I believe you have greater potential with that huge file
LennyBloke

Flan

Link Posted 11/01/2021 - 20:36
i did notice the slight shit of view and presumed it was on a tripod being indoors and mabye I'm trying to convince my self. will do some comparisons myself from a tripod with balanced lighting

RobL

Link Posted 11/01/2021 - 22:32
I donít think the subject has sufficient fine detail for a difference to be more obvious, you need something with a bit of texture - got any nuts left?

ronniemac

Link Posted 11/01/2021 - 23:27
Too close to call, and the slight differences I see might be due to variations in composition, camera shake, shake reduction movement, focus etc, or it might be that the movement which is detected causes the processor to discard parts of the raw files which are not aligned, effectively eliminating the pixel shift benefit. Best to use a tripod and set everything identical for both photos, manual focus and not altered between the two shots.

I cant believe that I have just opened two picture files and viewed them at max resolution wearing my glasses and looking through a magnifying glass.

LennyBloke

Link Posted 12/01/2021 - 10:03
Thanks for the comments, and I'm glad it has generated some interest - But I think we're agreed that a better test is needed - too many amateur elements in this quick comparison!
Also, it is important to clarify what is being tested.....

I am using Dynamic Pixel Shift (only available on the K1 MkII I believe) which means a tripod should not be used (it relies on the users natural slight movement to shift the frame). It is generally acknowledged that this feature is not as good as the standard Pixel Shift which benefits from using a tripod and is very precise system that moves the image by one pixel for each of 4 images - this will provide images with the greatest detail - but I hardly ever carry a tripod with me, so the fact that I can get a more detailed image when I'm out and about is a great plus (IMO)

Also, I am testing using JPEG only at this point - RAW can be used I believe - and the PS processing is done in the camera so I am working with a single Jpg file when post processing.

I will re-test using better subject matter (more texture, better lighting) and will try to match the composition better - it'll take a while in between work
LennyBloke

LennyBloke

Link Posted 12/01/2021 - 16:27
Here's another test from this afternoon - PP has zero sharpening and only highlights and shadows expanded to show fullest range, all settings the same - focus is on the centre of the hands of the pocket watch in the first pair and on the Queen Victoria stamp in the second pair (click for the bigger picture) ......

1. Without



2. With



3. Without



4. With



What do you think? Any significant (or subtle) differences?


LennyBloke

Chrism8

Link Posted 12/01/2021 - 17:24
Having seen these results this is something I think worth a try, I also agree with your choice of chocs, Mrs M8 and I are rather partial to the brand
Chris

www.chrismillsphotography.co.uk

" A Hangover is something that occupies the Head you neglected to use the night before".

-------------------------------------------------------------
K1 - Sigma 85mm F1.4, Pentax 150 -450 F4.5 / 5.6, Pentax FA 24 - 70 F2.8

Sigma 100-300 F4, Samyang 14mm F2.8, Sigma 70-200 F2.8,

K5iis - Sigma 17 - 70 F2.8, Sigma 70 - 300 F3.5/F5.6, Sigma 18 - 200 F3.6 / F4.5.

HarisF1

Link Posted 16/01/2021 - 09:54
Is it just me or do the colours seem slightly richer with DPS on?

LennyBloke

Link Posted 16/01/2021 - 10:54
HarisF1 wrote:
Is it just me or do the colours seem slightly richer with DPS on?

I have pixel-peeped these to death and can see very little fine detail differences (and some that I can see may well be due to slight position/focus changes) but I still think that the overall "look" has the edge with the DPS shots and it may well be the colour that does it

Even though I put quite a bit of effort into the "tests" i think that perhaps choosing a still-life subject for "Dynamic" pixel shift wasn't the best choice. Also, I'm using one of the finest Pentax macro lenses in its ideal focus distance - again probably not the best test!

I will persevere and take some comparison photos when I'm out and about and show them on here - the first few shots I took were on a dullish daytime walk and they did seem to have that little extra something so hopefully I'll be able to show that again but with "control" shots too
LennyBloke

Lubbyman

Link Posted 16/01/2021 - 11:28
LennyBloke wrote:
I have pixel-peeped these to death

The things we do to pass the time during lockdown...

Colours ought, in principle, to be more accurate because each pixel in the image is sampling R, G and B pixels from the sensor rather than interpolating from neighbours. What difference that makes in practice presumably depends on the subject. And since DPS depends on camera shake rather than precise movement of the sensor, there might also be a bit of luck involved in how the camera moved. Whatever, the effect does seem to be very subtle. Perhaps what is needed is an indication of what types of subject would most benefit from DPS.

And this all goes to show... that I'm not going to lose any sleep about not having a camera with pixel shift. My bank balance is grateful!

Steve
Add a Comment
You must be registered or logged-in to comment.