[normhead]

If you are interested in pixel shift for higher resolutions you can also increase the resolution with the D810 and D800 with photo stacking and also decrease noise



The one on the left is taken with pixel shift and the other on the right is using a photo stack using 4 shots. ISO 100, 200, 400 and iso 800 from the K1



here they are output to the same size. While it was not ideal using different iso ( I need 4 photographs for stacking and those are the only available) to stack images from you can see benefit from using stacking.

There are benefits to using stacking, with stacking it is beneficial if there is movement in between shots as this is how stacking works you need some movement. With pixel shift there can be no movement in between each exposure.
Another benefit is you can take as many frames as you want and pick the better of the lot to use in the stacking and ignore the ones you don't want to use.

And you are expecting stacking to be a built in feature on the Nikon? Like Pixel shift is built into the Pentax?

So you can stack with Nikon images.

I can shoot Pixel Shift and then stack on my Pentax.

Before you go there, I can do four shots where I bracket, Pixel shit, produce HDRs with the three files and then stack them, shooting with my K-1. K-1. And if I do that, I will be able to select the best files to stack. Better than D820 stacked files...

Honestly, don't you find this kind of reasoning to be kind of childish one-upmanship?

[Rondec]

Not sure how folks got to talking about pixel shift, but to me, the biggest benefit of pixel shift as Pentax does it is that it creates images with increased color depth and less noise and better dynamic range. When folks start talking super resolution, they are trying to create bigger images, but 36 megapixels is plenty for me.

Pentax does automate it and offers a version that compensates for subject movement, for what it's worth.

I sort of doubt that Nikon will include such a feature on their D810 sequel, but you never know.

[mee]

When that selection and improved AF performance allows you to use cheaper and lighter F/4 lenses and seeing a gain if performance helps in the pocket book at the end of the day.

Yes I completely agree with everything here. Couldn't have said it better.

A Practical Guide to Creating Superresolution Photos with Photoshop

I mainly use PhotoAcute PhotoAcute Studio Example

---------- Post added 02-03-2017 at 07:03 PM ----------

Here is the full image from the K1 stacked image



Its large 14720 pixels on the long side

You know, I saw that article years ago and tried to do recreate the effect but wasn't able to get similar result at the time. I'm using Affinity Photo now and somehow it works for me. I just trialed the technique by shooting a scene in my kitchen consisting of a bottle of peppercorns with my Sigma 105 macro.

I locked the zoom and the position in Infraview and cycle between the one of the individual frames and the SpR version. The upscaled version, downsized to the same size as the individual frames, does appear to have more detail in my simple test of 8 shots on burst.

A bit tedious but an interesting technique to me. Thanks again for reviving this for me.

PhotoAcute sadly appears to be long gone... last version was a couple of a years ago and no updates in their forum since then. Doesn't look good.

I'm rather amazed there aren't competing products on the market.. It seems the only thing PixelShift does on a Pentax camera is automate the process by taking 4 very distinct shots and merging together.

---------- Post added 02-03-17 at 08:26 PM ----------

Not sure how folks got to talking about pixel shift, but to me, the biggest benefit of pixel shift as Pentax does it is that it creates images with increased color depth and less noise and better dynamic range. When folks start talking super resolution, they are trying to create bigger images, but 36 megapixels is plenty for me.

Pentax does automate it and offers a version that compensates for subject movement, for what it's worth.

I sort of doubt that Nikon will include such a feature on their D810 sequel, but you never know.

Yes, it does seem to be much easier with a Pentax. But, correct me if I'm wrong, Ian was just mentioning that it is possible to get a pixel-shift like effect with the D810 or really any camera. It isn't Pentax exclusive. But the Pentax means does have it's pros.

I don't see there to be any way for a Nikon to automate the process since they don't have an IBIS system to move the sensor. I suppose they could install just a means of moving the sensor per pixel, but at that point why not just go all the way and offer IBIS? haha

The better option, I think, would be if they offered a NEF only version of PhotoAcute that was fully stocked with Nikon profiles and bundled it with some of their cameras.

The best option, would be if PhotoAcute came back with a new version that was fully stocked with way more profiles in general. Then we all win.

[Rondec]

I guess my point is that if you want a bigger image then Pentax's way isn't going to help you much. Pentax just gives you a better image that is the same size with regard to pixels as a normal K-1 image. It just happens to have extra DR and less noise. I can't imagine wanting to print bigger than I can with a K-1 pixel shift image but I suppose it could happen and that's where super resolution could help, I suppose or panoramas or any technique like that.

[mee]

I guess my point is that if you want a bigger image then Pentax's way isn't going to help you much. Pentax just gives you a better image that is the same size with regard to pixels as a normal K-1 image. It just happens to have extra DR and less noise. I can't imagine wanting to print bigger than I can with a K-1 pixel shift image but I suppose it could happen and that's where super resolution could help, I suppose or panoramas or any technique like that.

Oh I wasn't looking for a means to create bigger than what the K-1 can offer... just a means to do something similar on a Nikon body. This manual technique seems to do the trick..

[Rondec]

Oh I wasn't looking for a means to create bigger than what the K-1 can offer... just a means to do something similar on a Nikon body. This manual technique seems to do the trick..

But you can get the same effect by shooting with a longer lens and doing a panorama and there is a lot more software available that will merge those images. I think the question is how automated you want the process to be.

[mee]

But you can get the same effect by shooting with a longer lens and doing a panorama and there is a lot more software available that will merge those images. I think the question is how automated you want the process to be.

Panoramas are subject to their own set of unique issues.

I already have a working process, now, as shown by Ian. If PA is ever relaunched, it will become even quicker.

[Ian Stuart Forsyth]

PhotoAcute sadly appears to be long gone... last version was a couple of a years ago and no updates in their forum since then. Doesn't look good.

I don't know about this, the main page has changed and now says that they are working on a new version before they said nothing about new updates I have my fingers crossed as it was a pretty much automated function and even selected the best images from the bunch a couple clicks of the mouse and let the software do the work

---------- Post added 02-03-2017 at 11:25 PM ----------

Honestly, don't you find this kind of reasoning to be kind of childish one-upmanship?

No but I can clearly see it is to you

---------- Post added 02-03-2017 at 11:27 PM ----------

Before you go there, I can do four shots where I bracket, Pixel shit, produce HDRs with the three files and then stack them, shooting with my K-1. K-1. And if I do that, I will be able to select the best files to stack. Better than D820 stacked files...

After stacking 4-8 images you wont see that much of a difference

---------- Post added 02-03-2017 at 11:30 PM ----------

Yes, it does seem to be much easier with a Pentax. But, correct me if I'm wrong, Ian was just mentioning that it is possible to get a pixel-shift like effect with the D810 or really any camera. It isn't Pentax exclusive. But the Pentax means does have it's pros.

To see a real increase in resolution you need sub pixel movement this is one of the reasons why you do see an increase with photo stacking at about 8 - 10 stack is the best but even using 4 as shown in the K1 photo you can see a difference. This was shown to me by a fellow shooter that uses another brand of camera that also uses pixel shifting tech and felt that the use of photo stack was his choice of decreasing noise ( increase DR) and increase resolution over what his camera could do.

As he had pointed out to me that it is much hard to have the camera take 4 or 8( 8 for the camera he uses) images and not have the camera change positions (with 36 mp its even harder) Only using this for a short time I can see why he uses this technique over his cameras sensor shifting.
The shoot taken of the Lacombe main street there was conditions that would impede the use of pixel shifting
The stacking was taken using 4 sec exposures giving a total time 16sec, traffic really limited the time where I could take the photo and had several where someone turned off of a side street wreaking that exposure and when that happened I could just delete that image and take another. With pixel shift you are held to 16 sec ( plus the time between 4 exposures ) to get your capture

[cali92rs]

Not sure how folks got to talking about pixel shift, but to me, the biggest benefit of pixel shift as Pentax does it is that it creates images with increased color depth and less noise and better dynamic range. When folks start talking super resolution, they are trying to create bigger images, but 36 megapixels is plenty for me.

Pentax does automate it and offers a version that compensates for subject movement, for what it's worth.

I sort of doubt that Nikon will include such a feature on their D810 sequel, but you never know.

The super resolution that Ian is talking about also decreases noise. Fwiw, i don't think either method increases dynamic range. They both increase signal to noise ratio (due to decrease in noise), which allows you to pull out shadows with less noise. But that is different than increasing dynamic range.
DCShooter explains it better than I can:

Question about Pentax's version of Pixel Shift Resolution - PentaxForums.com

[awscreo]

Interesting thread. To me k-1 is an almost perfect camera, for its price point. The only thing I sometimes struggle with is the af system, and that could be due to my lenses (I don't own any DFA ones, hopefully picking up the 28-105 soon, will see if af is better with it). I rarely use pixel shift right now, hoping to implement it into my workflow more come warmer season when I will have more time and easier access to landscapes. I used it few times, and was impressed with the image quality I got. Whatever d810 successor will be would definitely be an amazing camera. I think the whole dxxx lineup holds up really well IQ wise. I was really considering d750 when I was looking for an upgrade, it's a real work horse of a camera, but overall I'm happy I ended up with k-1, I think I will not be upgrading for a good while.

[Ian Stuart Forsyth]

The super resolution that Ian is talking about also decreases noise. Fwiw, i don't think either method increases dynamic range. They both increase signal to noise ratio (due to decrease in noise), which allows you to pull out shadows with less noise. But that is different than increasing dynamic range.

The DR is all based on what is an acceptable level of noise, lets take it to the extreme and say that any shot at iso 100 pushed to 400 is an in the max acceptable level of noise you would tolerate rather than what DXO deems as acceptable level of noise. If we decrease that noise in the image by one stop by using more data then that noise at iso 100 pushed to 800 will show the same amount of noise as the 400 push. So now with that increased data I have accessed to more DR with the same level of noise.

[biz-engineer]

The DR is all based on what is an acceptable level of noise, lets take it to the extreme and say that any shot at iso 100 pushed to 400 is an in the max acceptable level of noise you would tolerate rather than what DXO deems as acceptable level of noise. If we decrease that noise in the image by one stop by using more data then that noise at iso 100 pushed to 800 will show the same amount of noise as the 400 push. So now with that increased data I have accessed to more DR with the same level of noise.

Stacking is limited, at a certain point, decreasing noise levels does not increase dynamics range at all, because the 14 bits A/D can't code more than 14bits.
At ISO100, the noise is already low, 4 stacked shots decrease read noise essentially, beyond that 4x averaging of the Pentax pixel shift, there is no gain of DR.
Beyond averaging, the only way to increase dynamic range is exposure bracketing.

[Rondec]

The super resolution that Ian is talking about also decreases noise. Fwiw, i don't think either method increases dynamic range. They both increase signal to noise ratio (due to decrease in noise), which allows you to pull out shadows with less noise. But that is different than increasing dynamic range.
DCShooter explains it better than I can:

Question about Pentax's version of Pixel Shift Resolution - PentaxForums.com

By giving less noise in the shadows you essentially have a little more dynamic range. If you look at DXO One camera versus the DXO One SuperRAW (the only super resolution mode that DXO Mark has tested), you find that there is some improvement in dynamic range but more improvement in acceptable high iso level.

I just find that I can take a pixel shift image from the K-1 and push it a lot harder with noise showing up and the image looking stretched than I can push a normal iso 100 image. That's why I use it for many landscape images. Yes, I could just take four or five images and find a third party software to combine them, but it wouldn't be as easy. The automation of the process is what makes it worthwhile to me.

[normhead]

By giving less noise in the shadows you essentially have a little more dynamic range. If you look at DXO One camera versus the DXO One SuperRAW (the only super resolution mode that DXO Mark has tested), you find that there is some improvement in dynamic range but more improvement in acceptable high iso level.

I just find that I can take a pixel shift image from the K-1 and push it a lot harder with noise showing up and the image looking stretched than I can push a normal iso 100 image. That's why I use it for many landscape images. Yes, I could just take four or five images and find a third party software to combine them, but it wouldn't be as easy. The automation of the process is what makes it worthwhile to me.

Exactly....
If I wanted to spend a lot of time stacking photos, I'd be doing it. But the simple fact is, for most images, i won't stack and I won't stitch. But what I find really funny about this, is when the D800, some Pentax users said if we want image files that big we can stitch. D800 users like sherman accused us of inventing work arounds to make up for the deficiencies of our cameras.

It would seem, now the shoes on the other foot, and we have Nikon users suggesting work arounds. How things have changed. I guess it's only a bad thing when Pentax users do it.

I detect a bit of camera preference prejudice taking place here. Prejudice in the sense of pre-judging. NO facts will be accepted unless Nikon comes out ahead of Pentax. And if Pentax has an advantage, Nikon users can work around it. I see how this works.

BY the way, there's a word for it, duplicity.

It's when different standards are applied differently to different but similar things, based on one's prejudices.

Personally, I rarely bother to stitch, I've never bothered to stack, but so far I have over 50 Pixel Shifted images. Based on my experience, I'd say the difference between Pixel shifting and stacking is the difference between doing or not doing. And that is a very important difference.

[mee]

Stacking is limited, at a certain point, decreasing noise levels does not increase dynamics range at all, because the 14 bits A/D can't code more than 14bits.
At ISO100, the noise is already low, 4 stacked shots decrease read noise essentially, beyond that 4x averaging of the Pentax pixel shift, there is no gain of DR.
Beyond averaging, the only way to increase dynamic range is exposure bracketing.

Maybe there are two types of stacking and thus two types of dynamic range 'increase' on going in this discussion?

One way is shooting multiple bursts at multiple exposure levels (using the super resolution burst mode method) and then stacking them. The benefit is an 'hdr' like image with more leeway than a single shot, plus increased detail.

The other way of going about is not about generating more DR than at the minimum ISO, but approaching that level at a higher ISO setting more cleanly. This is done by simple shooting a (high iso) burst at one exposure level and stacking. You do get a cleaner resultant image, that is more noticeable at higher ISO, and lets one play with raising shadows a bit more freely.