[Sandy Hancock]

Autofocus... I think is the answer to that.

Nope. F and FA lenses are autofocus and still tiny. Silent in-lens AF motors take up a bit more space but that’s not the main issue.

The difference is the modern emphasis on highly corrected optics to negate chromatic aberration and provide optimal sharpness at wide apertures.

The good news is that all old lenses still work, so our choices are expanding

[pschlute]

Lighter camera/lens assembly is easier to hold steady for a lot of people.

You dont want to hold it too steady or you will negate the Dynamic Pixel Shift effect

[amoringello]

I'm wondering what percentage of people can hold the camera steady enough for the SR to work at a less than 1-pixel resolution for a one second hand held shot.
Even with very high shutter speed, the total process to take four images is nearly four seconds.
I can get sometimes useable images at one second exposure (compared to not using SR), but I'd hardly let someone view them full size... i.e. the sensor is not able to make a perfectly steady image over that time range.
This just seems like snake-oil to me, and I don't see where a heaver or lighter lens isn't going to help much.

[RobA_Oz]

I'm wondering what percentage of people can hold the camera steady enough for the SR to work at a less than 1-pixel resolution for a one second hand held shot.
Even with very high shutter speed, the total process to take four images is nearly four seconds.
I can get sometimes useable images at one second exposure (compared to not using SR), but I'd hardly let someone view them full size... i.e. the sensor is not able to make a perfectly steady image over that time range.
This just seems like snake-oil to me, and I don't see where a heaver or lighter lens isn't going to help much.

We're all waiting to see how it works in reality, but the 1 second total time claim has already been dissected and dismissed as implausible.

[Gimbal]

Even with very high shutter speed, the total process to take four images is nearly four seconds.

What? How do you figure that?

[Rondec]

I'm wondering what percentage of people can hold the camera steady enough for the SR to work at a less than 1-pixel resolution for a one second hand held shot.
Even with very high shutter speed, the total process to take four images is nearly four seconds.
I can get sometimes useable images at one second exposure (compared to not using SR), but I'd hardly let someone view them full size... i.e. the sensor is not able to make a perfectly steady image over that time range.
This just seems like snake-oil to me, and I don't see where a heaver or lighter lens isn't going to help much.

I don't think it takes that long to actually shoot the images, although the processing does take a bit of time after the images are a shot, you don't have to hold the camera still during that time period.

I think the point with this hand held pixel shift is that the camera is not still between shots. Rather the camera is going to do some type of auto alignment in camera and fill in details between images from the other three images shot. Pentax says it will be 70 percent as effective as traditional pixel shift, which is still pretty decent. My guess is that it will reduce noise quite a bit, but not improve resolution quite as much.

There are guides to doing this type of super resolution shooting outside of the camera.

A Practical Guide to Creating Superresolution Photos with Photoshop

[RobA_Oz]

Nope. F and FA lenses are autofocus and still tiny. Silent in-lens AF motors take up a bit more space but that’s not the main issue.

The difference is the modern emphasis on highly corrected optics to negate chromatic aberration and provide optimal sharpness at wide apertures.

The good news is that all old lenses still work, so our choices are expanding

The new D FA50 looks like it may be internal focussing, which, of course helps with designing weather resistance. However, if so, I imagine that will add to complexity and bulk – the DA*50-135 seems to me to be an example of that.

[pschlute]

I believe Rondec is correct. I think the point of DPS is not that each image is captured with 1 pixel shifted, but that the natural movement of the hands, corrected by the shake reduction system, will give you 4 images each with pixels shifted by the movement of the camera. They will be shifted by a number of pixels randomly, not one pixel. The software must then compare the 4 images and align them to produce the final image.

This would mean that there is no guarantee that each pixel will get the same 1xR; 2xG; 1xB colour exposure, and I guess this is where the 70% effective figure comes from.

This is how I think it will work at any rate.

[Kunzite]

What? How do you figure that?

Because the sensor can't transfer the data any faster.

[Gimbal]

Because the sensor can't transfer the data any faster.

The camera can transfer 4.4 images per second, so why should it take 4 seconds to take a dynamic PS (four images)? One second is more like it.

[Kunzite]

Ouch, my bad! I completely misread that part, thought it said one second.
You're right, of course.

[Rickster]

Inertia (which in turn is caused by mass) actually helps the other way round.

This!!!

---------- Post added 02-28-18 at 01:22 PM ----------

I believe Rondec is correct. I think the point of DPS is not that each image is captured with 1 pixel shifted, but that the natural movement of the hands, corrected by the shake reduction system, will give you 4 images each with pixels shifted by the movement of the camera. They will be shifted by a number of pixels randomly, not one pixel. The software must then compare the 4 images and align them to produce the final image.

This would mean that there is no guarantee that each pixel will get the same 1xR; 2xG; 1xB colour exposure, and I guess this is where the 70% effective figure comes from.

This is how I think it will work at any rate.

I'll go on record as saying no to this speculation and say I think all the camera needs to do is add/subtract the appropriate pixel offsets to the shake reduction solution ... at least that's the way I would attempt it. ;-) This presumes that the shake reduction can deliver a pixel accurate solution, 70% of the time.

[pschlute]

I'll go on record as saying no to this speculation and say I think all the camera needs to do is add/subtract the appropriate pixel offsets to the shake reduction solution ... at least that's the way I would attempt it. ;-) This presumes that the shake reduction can deliver a pixel accurate solution, 70% of the time.

I am not sure i follow your line of thought here ?

[Oakland Rob]

From the front page here, with a hands-on with Pentax and the K-1ii at WPPI:

We talked about the new K-1 II's dynamic pixel shift resolution mode with Dan Savoie, and we were able to test it with the 50mm. Previously, the K-1 could use its pixel shift technology to increase the resolution and color accuracy of images (of still subjects) when shooting on a tripod. The new system allows for hand-held shooting. It uses the mechanical shutter in-between shots (compared to the electronic shutter on standard pixel shift). In hand it feels like exposure bracketing hand-held with in-camera image alignment. It appeared to work well at the show, but of course we can't draw any final conclusions at this point.

Read more at: Ricoh Imaging Exhibits Pentax at WPPI 2018 - Trade Shows | PentaxForums.com

[monochrome]

I believe Rondec is correct. I think the point of DPS is not that each image is captured with 1 pixel shifted, but that the natural movement of the hands, corrected by the shake reduction system, will give you 4 images each with pixels shifted by the movement of the camera. They will be shifted by a number of pixels randomly, not one pixel. The software must then compare the 4 images and align them to produce the final image.

This would mean that there is no guarantee that each pixel will get the same 1xR; 2xG; 1xB colour exposure, and I guess this is where the 70% effective figure comes from.

This is how I think it will work at any rate.

The IBIS System captures all the natural motions of the 4 hand held images in the 5 axes. The camera software - perhaps using the Accelerator Unit - aligns and combines the 4 images using the motion data captured by the IBIS system. It isn’t PixelShift - it is Enhanced Resolution created in-camera, using the external motion info to create the enhanced (SuperResolution) Image. The enhancement process occurs after the 4 images were captured, so they could presumably be captured at the absolute fastest fps the camera can handle.