Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
04-28-2018, 08:54 PM   #1
Site Supporter
Site Supporter
BruceBanner's Avatar

Join Date: Dec 2015
Location: Blue Mountains, NSW, Australia
Posts: 2,679
Creating Super Resolution Images Handheld (like a K-1mkii, but not).

Now that the K-1mkii is out there's been a lot of chitchat about the Dynamic Pixel Shift feature. I came across a few interesting stories on how creating Super Resolution photos is nothing new, it's something I've never tried seeing as I always did pixel shift on a tripod with the K-1, but this one article in particular made my ears prick up;

Here's how to create a super resolution photo with any camera: Digital Photography Review

What really stood out for me was that the technique used was handheld (if I am not mistaken). With regular pixelshift on the k1 there is a sensor bumping going on, with the k1ii there is the SR feedback being used to assist, and I guess this technique here also requires some movement by the user in order to create the pixelshifting, it's just that it's handled all within photoshop?

So basically, is this another reason for a K-1 user not to go the upgrade route (if specifically seeking out handheld pixelshifting), that is.. in a way we already have that feature if we just know how to go about obtaining it?

I was intrigued to see also that the photoshop route looked significantly superior to pixelshifting. Furthermore... could we obtain even better results by doing a bracketing/hdr (if not overly abusing the exposure changes)?

Currently I have not read nor understood fully how to take the shots and process them properly, I've skim read and have another article about the topic to read. Little makes sense to me until I attempt it, stuff it up, try again etc. (it's not uncommon for me to take weeks to learn something however I just wanted to put this out there to perhaps more knowledgeable chaps that may even already be doing this 'trick'. Perhaps their views can provide more insights into this technique.

Some further reading info;

A Practical Guide to Creating Superresolution Photos with Photoshop

Pentax K-1: an open-source photo-processing workflow - Stream of Consciousness

Cheers,

Bruce

04-28-2018, 09:28 PM   #2
maw
Pentaxian
maw's Avatar

Join Date: Feb 2014
Location: Sassari (Italy)
Posts: 465
QuoteOriginally posted by BruceBanner Quote
I came across a few interesting stories on how creating Super Resolution photos
I have known the technique of Ian Norman for 3 years or more, the idea is excellent, but not so much as you can get other special effects, but in contrast the processing is long,

you also need a computer powerful enough.

You have to take a lot of shots (I don't remember exactly), but at least twenty I think. It is preferable to use the pixel shifting in camera.

Forget if there is a moving object you get an effect similar to this.


Last edited by maw; 04-29-2018 at 03:57 AM.
04-28-2018, 09:29 PM   #3
Administrator
Site Webmaster
Adam's Avatar

Join Date: Sep 2006
Location: Arizona
Photos: Gallery | Albums
Posts: 47,545
It's kind of similar, but different, because traditional super resolution techniques through post-processing seek to mitigate bayer interpolation by combining multiple exposures to reduce noise and sharpen details, whereas the Pentax PSR actually calculates the missing color values with a high degree of certainty.

Given the fact that hand-held PSR is so easy to use, I wouldn't say that it is rendered obsolete by this technique. Plus, if you wanted the absolute best image quality, you'd be bringing a tripod in any case.

Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com server and development costs are user-supported. You can help cover these costs by donating. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

04-28-2018, 09:41 PM   #4
Site Supporter
Site Supporter
BruceBanner's Avatar

Join Date: Dec 2015
Location: Blue Mountains, NSW, Australia
Posts: 2,679
Original Poster
QuoteOriginally posted by maw Quote
I have known the technique of Ian Norman for 3 years or more, the idea is excellent, but not so much as you can get other special effects, but in contrast the processing is long,

you also need a computer powerful enough.

You have to take a lot of shots (I don't remember exactly), but at least twenty I think. It is preferable to use the pixel shifting in the room.

Forget if there is a moving object you get an effect similar to this.
In the first article I linked (dpreview) it did show comparisons between 4 and 20 images, stating that there were diminishing returns the more you add (basically 4 is enough).

You get issues with moving objects with PSR as well... Rawtherapee helps a lot with that, which makes me wonder... can you do the Ian Norman technique and then take off to RT for some motion correction?

QuoteOriginally posted by Adam Quote
It's kind of similar, but different, because traditional super resolution techniques through post-processing seek to mitigate bayer interpolation by combining multiple exposures to reduce noise and sharpen details, whereas the Pentax PSR actually calculates the missing color values with a high degree of certainty.

Given the fact that hand-held PSR is so easy to use, I wouldn't say that it is rendered obsolete by this technique. Plus, if you wanted the absolute best image quality, you'd be bringing a tripod in any case.
The article pictures however make PSR look less than stella compared with Ian Norman's technique, that's the part that interests me. I mean I tend to always have a tripod with me on bushwalks, for slo mo waterfalls and the like. But maybe sometimes it would be nice to take a SRI handheld (but I only have a K-1), there's been a couple of landscape points where the tripod just did not work, handheld would have worked a lot better.

04-28-2018, 09:43 PM   #5
maw
Pentaxian
maw's Avatar

Join Date: Feb 2014
Location: Sassari (Italy)
Posts: 465
QuoteOriginally posted by Adam Quote
It's kind of similar, but different, because traditional super resolution techniques through post-processing seek to mitigate bayer interpolation by combining multiple exposures to reduce noise and sharpen details, whereas the Pentax PSR actually calculates the missing color values with a high degree of certainty.
Yes, correct Adam.
04-29-2018, 12:19 AM   #6
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 31,018
QuoteOriginally posted by BruceBanner Quote
In the first article I linked (dpreview) it did show comparisons between 4 and 20 images, stating that there were diminishing returns the more you add (basically 4 is enough).
Ian Norman says a minimum of 20 exposures as a rule of thumb and a close read of the DPR article does not contradict that. It depends on how the merge is managed and the intent of the shot. Assuming you have Photoshop or other suitable raster editor (Affinity Photo will work), why don't you give it a try and post comparative results. After all, you have the K-1.


Steve
04-29-2018, 12:30 AM   #7
Site Supporter
Site Supporter
BruceBanner's Avatar

Join Date: Dec 2015
Location: Blue Mountains, NSW, Australia
Posts: 2,679
Original Poster
QuoteOriginally posted by stevebrot Quote
Ian Norman says a minimum of 20 exposures as a rule of thumb and a close read of the DPR article does not contradict that. It depends on how the merge is managed and the intent of the shot. Assuming you have Photoshop or other suitable raster editor (Affinity Photo will work), why don't you give it a try and post comparative results. After all, you have the K-1.


Steve
I definitely aim to do that Steve, when time allows. I just wanted this topic put out there to see if any of our users actually do this (or a process like it) already. I'm also not a photoshop pro so I am wary to getting the right results, from start to scratch. For example, even taking the shot itself, what mode, should we just do a continuous burst perhaps up till the buffer kicks in etc?

"the difference between the 4 image and 20 image super resolution examples is less noticeable. While there is some advantage to stacking more images, returns are diminishing in this case."

I don't know about your eyes but mine tell me there isn't much to be gained over 20 images vs 4, especially when we're talking the extra work involved to take and process 20 vs 4. I already see a clear distinction between 4 median or average vs PSR, that's enough of a difference for me to make this process seem interesting and worthwhile investigating further. (not to mention 20 takes longer to take, therefore motion blurring increases)
04-30-2018, 05:47 AM   #8
Site Supporter
Site Supporter
BruceBanner's Avatar

Join Date: Dec 2015
Location: Blue Mountains, NSW, Australia
Posts: 2,679
Original Poster
Ok guys, I just tried a stab at making my first Super Resolution photo, handheld and with my K-1 (just a K-1, not a K-1mkii).

I ran into some hiccups along the way, I basically followed the procedure outlined here;

A Practical Guide to Creating Superresolution Photos with Photoshop

I took 20 shots but I seriously could not process the pictures, certain stages of the conversion process just had my entire system locked up for 30mins easily with no end in sight (I have a feeling the progress bar is unreliable perhaps staying at a 10th of the way but when in fact could be close to completion). Either way I gave up on 20 shots and instead worked on just 4.

Even doing 4 shots had some serious waiting time for certain stages to complete. I would guess on my system I was waiting about 20-30mins from start to finish to get the final SR file. I'll get more accurate specs for you later but I am only running my system on an Intel NUC Skull Canyon, core i7 and 16gb RAM. But at least with 4 pictures and PS processing certain changes, it wasn't a complete lock up of the system, I could still browse the web etc. with 20 I felt it was hard to navigate around the desktop.

I chose to do the Smart Object route at point 4 of the guide, and process the stack as Median. I also did the sharpness filter with the recommended settings in the article.

Below are two links to the files as exported jpgs, no processing has been done to the pictures in LR, no exposure changes or anything. The jpg file ending 'RAW' is the last frame of the 4 shots taken converted simply to jpg, the jpg file ending SR is the super resolution RAW file exported in LR as jpg, again no further processing done other than the filter sharpening in PS.

I have also included the RAW file of the final fourth shot used so you can boot that up and see if you can tweak it to get somewhere close to the SR file. What I was discovering was increasing sharpness wasn't overly helping much, it brought in way too much noise and actually still never caught up with the sharpness the SR file had :O

You'll see by the EXIF but I tried to stay within the recommended parameters, but it was late in the day and I was losing light so not quite ideal, however I quite like the image as there is nice light and darker places to inspect.

EXIF is basically;
K-1
DA40mm XS
f5.6
1/40th
ISO 100

(and I can verify each of the 4 files used were all the same in that respect)

IMGP8222RAW.Jpg

IMGP8222SR.Jpg

IMGP8222.DNG

One weird thing that did occur. After I did all the processing in PS and went to 'Save As' (a .TIFF), the file literally was too large to save, it was 3.9gb (lol) and I got an error message saying I couldn't save files larger than 4gb. I scratched my head not knowing what to do next, so I went quickly to Image Size and selected 50% and tried again, and this made the TIFF file save as a mere 2.92gb hehe. I guess I am not fully understanding the resizing step of the guide, I must be missing something because with 20 files it must be astronomical file to try and save/export! Surely with 4 files we're looking at an exported TIFF or around the same size as PSR file (100-200mb).

Either way I am actually really happy with the results. From my understanding using the Dynamic Pixel Shift shot of the K-1mkii really stops the camera in it's tracks. It's a reasonable wait time till the camera has finished processing and you can resume shooting. That really doesn't sit too well with me to be honest, I'd far rather this route, a quick 4 files fired off in rapid succession and a longer processing time at home whilst you have some toast and tea =)

Next up I will try some HDR done in this fashion, I am intrigued to see how the camera copes with a proper handheld landscape 5 bracketed shot done with this method.

Lemme know what you all think!

Cheers,

Bruce

04-30-2018, 09:29 AM   #9
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 31,018
QuoteOriginally posted by BruceBanner Quote
I took 20 shots but I seriously could not process the pictures, certain stages of the conversion process just had my entire system locked up for 30mins easily with no end in sight
Oh...yes, merging 20 36Mpx layers could be time consuming


Steve
04-30-2018, 09:32 AM   #10
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 31,018
QuoteOriginally posted by BruceBanner Quote
Either way I am actually really happy with the results.
Yes, they look good.

Steve
04-30-2018, 10:46 AM   #11
maw
Pentaxian
maw's Avatar

Join Date: Feb 2014
Location: Sassari (Italy)
Posts: 465
QuoteOriginally posted by BruceBanner Quote
Ok guys, I just tried a stab at making my first Super Resolution photo, handheld and with my K-1 (just a K-1, not a K-1mkii).
I wrote before that it's a long, boring and heavy technique, it's definitely more convenient to use Pixel Shifting, it's done in seconds, and then you can open it in Raw Theraphee
by choosing which level to save in TIFF or other format.

1. Remember also that in PS you have these maximum values that you cannot exceed.

2. Photoshop supports documents up to 300,000 pixels in either dimension and offers three file formats for saving documents with images having more than 30,000 pixels in either dimension.
Keep in mind that most other applications, including versions of Photoshop earlier than Photoshop CS, cannot handle files larger than 2 GB or images exceeding 30,000 pixels in either dimension.
Choose File > Save As, and choose one of the following file formats: Large Document Format (PSB)

3. Supports documents of any file size. All Photoshop features are preserved in PSB files (though some plug-in filters are unavailable if documents exceed 30,000 pixels in width or height).
Currently, PSB files are supported only by Photoshop CS and later.

4. Photoshop Raw Supports documents of any pixel dimension or file size, but does not support layers. Large documents saved in the Photoshop Raw format are flattened.
TIFF Supports files up to 4 GB in size. Documents larger than 4 GB cannot be saved in TIFF format. That's why you can't save documents over 4 GB in PS.

Also here is the link that explains exactly the procedure Pixel Shift
04-30-2018, 11:23 AM   #12
Site Supporter
Site Supporter




Join Date: Dec 2016
Location: Ulfborg, Denmark
Photos: Albums
Posts: 266
T&C shows how to beat a D810 IQ wise with a K-200:
It takes some more post processing to get there, but I guess knowing you have saved 3900$ will ease the “pain”
04-30-2018, 01:52 PM   #13
Site Supporter
Site Supporter
BruceBanner's Avatar

Join Date: Dec 2015
Location: Blue Mountains, NSW, Australia
Posts: 2,679
Original Poster
QuoteOriginally posted by maw Quote
I wrote before that it's a long, boring and heavy technique, it's definitely more convenient to use Pixel Shifting, it's done in seconds, and then you can open it in Raw Theraphee
by choosing which level to save in TIFF or other format.

1. Remember also that in PS you have these maximum values that you cannot exceed.

2. Photoshop supports documents up to 300,000 pixels in either dimension and offers three file formats for saving documents with images having more than 30,000 pixels in either dimension.
Keep in mind that most other applications, including versions of Photoshop earlier than Photoshop CS, cannot handle files larger than 2 GB or images exceeding 30,000 pixels in either dimension.
Choose File > Save As, and choose one of the following file formats: Large Document Format (PSB)

3. Supports documents of any file size. All Photoshop features are preserved in PSB files (though some plug-in filters are unavailable if documents exceed 30,000 pixels in width or height).
Currently, PSB files are supported only by Photoshop CS and later.

4. Photoshop Raw Supports documents of any pixel dimension or file size, but does not support layers. Large documents saved in the Photoshop Raw format are flattened.
TIFF Supports files up to 4 GB in size. Documents larger than 4 GB cannot be saved in TIFF format. That's why you can't save documents over 4 GB in PS.

Also here is the link that explains exactly the procedure Pixel Shift
Thanks for that maw. So you seem like a knowledgeable chap, can you then assist with the process outlined by Ian Norman, clearly something is going wrong with his steps if I am taking only a mere 4 RAW images and unable to save as I am coming out with a file larger than 4gb. Resizing at the end by 50% to get the file down to 2.92gb was a stab in the dark...
Perhaps this article is old, and when Ian wrote it it was not intended for 36mp RAW files but files and resolution for smaller? Should the Image Resize step no.2 just be omitted completely? Or set to a different % from the start? I don't mind generating a large 2-4gb tiff file, as I noticed that LR seems ok to fire up my 2.9gb file for editing and then exporting with everything set to Jpg max it turns into a 20mb pic
But perhaps there is a way to make the tiff file generated from this process to be something lower without losing quality?

Cheers,

Bruce
04-30-2018, 03:09 PM   #14
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 31,018
QuoteOriginally posted by BruceBanner Quote
Perhaps this article is old, and when Ian wrote it it was not intended for 36mp RAW files but files and resolution for smaller?
He was working from 24Mpx (Sony A7II) captures in 2015 and with the initial upsize, each element of the stack accounted for 94Mpx, though after flattening, the final image would still be only 94Mpx for a JPEG of only about 14MB on disk. (At least that is the size of the downloaded full image file.) Speaking of the downloaded file, the 94Mpx at full resolution is not very impressive.


Steve
04-30-2018, 05:44 PM   #15
maw
Pentaxian
maw's Avatar

Join Date: Feb 2014
Location: Sassari (Italy)
Posts: 465
QuoteOriginally posted by BruceBanner Quote
Thanks for that maw. So you seem like a knowledgeable chap, can you then assist with the process outlined by Ian Norman, clearly something is going wrong with his steps if I am taking only a mere 4 RAW images and unable to save as I am coming out with a file larger than 4gb. Resizing at the end by 50% to get the file down to 2.92gb was a stab in the dark...
Perhaps this article is old, and when Ian wrote it it was not intended for 36mp RAW files but files and resolution for smaller? Should the Image Resize step no.2 just be omitted completely? Or set to a different % from the start? I don't mind generating a large 2-4gb tiff file, as I noticed that LR seems ok to fire up my 2.9gb file for editing and then exporting with everything set to Jpg max it turns into a 20mb pic
But perhaps there is a way to make the tiff file generated from this process to be something lower without losing quality?

Dear Bruce,

In the article Ian Norman explains step by step the procedure that looks at random quite well, because I had seen a tutorial of Deke McClelland different in content but similar in form, maybe one day I will publish it "if I have time.

So the solution has two faces, in my opinion, "Ian says nothing about how to save the image assuming he saves it to PSD or JPEG.

The first solution is to resize the files originally "Ian uses a 24 mpx camera", so you should first resample the images of the Pentax K1 (36 to 24 mpx) to get a smaller final file (about half).
Or do it in the end generally the best is the Bicubic Resampling.

Because as mentioned above Tiff files cannot exceed 4GB. Otherwise you can save them in another Jpeg, PSD, EPS, BMP format. PS "Ipse dixit".

Below are the specifications of the TIFF resolution


Tagged-Image File Format (TIFF, TIF) is used to exchange files between applications and computer platforms. TIFF is a flexible bitmap image format supported by virtually all paint, image-editing,
and page-layout applications. Also, virtually all desktop scanners can produce TIFF images. TIFF documents have a maximum file size of 4 GB.

TIFF format supports CMYK, RGB, Lab, Indexed Color, and Grayscale images with alpha channels and Bitmap mode images without alpha channels. Photoshop can save layers in a TIFF file;
however, if you open the file in another application, only the flattened image is visible. Photoshop can also save notes, transparency, and multiresolution pyramid data in TIFF format.

In Photoshop, TIFF image files have a bit depth of 8, 16, or 32 bits per channel. You can save HDR images as 32-bits-per-channel TIFF files.

Cheers, Mario
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, files, images, k-1, photography, photos, photoshop, pixel, post, process, resolution, resolution images, route, stack, technique, time, user, vs
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
SLR Lounge- Noise comparison shot K-1 vs K-1mkII @ISO 12800 - good improvement interested_observer Pentax News and Rumors 51 03-06-2018 11:42 AM
SR II on K-1 MKII will offer Dynamic (handheld) Pixel Shift Resolution alpheios Pentax News and Rumors 84 02-22-2018 02:21 PM
Not one, not two, not three, not four, but a wedding where half attendees are bob. LeDave Photographic Industry and Professionals 12 05-16-2016 03:40 AM
Banker welfare and why we're not creating jobs Nesster General Talk 11 10-06-2010 04:56 PM
storing images, creating a usable library? landscaped1 Troubleshooting and Beginner Help 10 01-28-2010 05:45 PM



All times are GMT -7. The time now is 04:48 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top