[MJKoski]

Now that different options have been discussed, here are examples of the same factory scene as above with Sony A7R1, R2 and Canon 90mm TS-E.

Pick your poison:

1) Canon 90mm TS-E, maximum swing:


2) Canon 90mm TS-E, almost maximum forward tilt (note thin wedge of focus in vertical direction):


3) Canon 90mm TS-E all settings zeroed, focus stack with aperture bracketing (foreground poles)


Helicon Focus did not manage to blend the poles with the background until I used one exposure with well stopped down aperture and it still is not perfect but acceptable.

[BigMackCam]

Well, 40x 60" posters framed on canvas shot with 22Mpixels are easy to find at Ikea and Amazon, you can have one for 39 euros or less. No need to go through the effort of using a Pentax camera and going through the creation of the photographs, post processing , ordering it to a photolab and framing it. However, the prints shot on Phase One that are sold by Lumas in Berlin, Paris and Vienna sell at prices from 3500 euros to 8000 euros each. The price for a 40 x 60" C-print on high end paper (from the same lab as the one working for Lumas) costs me between 100 and 200 euros, so you can see my stitching and stacking effort could be worth several thousand euros.

That context - your motive, essentially - goes some way in explaining why you're prepared to go to so much trouble. Lofty heights, indeed! Had you mentioned this at the outset, I suspect you might have avoided some of the earlier critical responses...

[RICHARD L.]

Ansel Adams was using an 8X10 view camera with a 300 mm normal lens when he began making landscapes. He climbed on a platform atop his Cadillac station wagon to put distance between his lens and his foreground elements, using the Scheimpflug Principle to bring every important focal planes into sharp focus, letting small apertures like f/32, f/45 and even f/64 take care of the middle blurry parts of the image.


Regards

[StiffLegged]

You don’t think Ansel did that to shoot over irritating foreground details like fences, road verges, people? I don’t think DOF was the primary concern, climbing 6 feet off the ground lifts your viewpoint up so you get a more expansive view of fore and middle ground. A fine example is “Mt Williamson from Manzanar.”

Not to be dismissive, but people pay thousands of euros for big prints of fine images, not just because they were shot on Expensive Camera X. But if you have images that size and customers willing, then go for it.

[fehknt]

I think to address (partially) your issues with stacked-panos: have you tried super-resolution stacked (non-pano)? I.e., something like 12 focus stacked images with the same lens and the same framing, but with a minor vibration applied to the tripod so that it shifts around some non-integer number of pixels. It'll be a ton of data but should enable higher resolution focus stacked results without the panorama stitching issues. (Obviously, enabling pixel-shift on the stack is a good first step, but it's not entirely a replacement for super resolution as far as spatial resolution is concerned.)

[MJKoski]

I use PS a lot when exposing for focus stacks. Helicon Focus seems to be able to take advantage of added details depending on stacking mode. Method B and C produce less errors.

[biz-engineer]

Not to be dismissive, but people pay thousands of euros for big prints of fine images, not just because they were shot on Expensive Camera X.

Sure. That's why I'm trying to bump quality by using a common camera and elaborating a workflow.

[biz-engineer]

I think to address (partially) your issues with stacked-panos: have you tried super-resolution stacked (non-pano)? I.e., something like 12 focus stacked images with the same lens and the same framing

I haven't tried that yet. Being able to leave the camera at the same framing has the advantage that the composition you get is what you see, whereas pano stitching often requires cropping.

I have to try and compare three methods:

- focus stacking with superres or PS, keeping camera in place
- aperture bracketing with superres / PS, keeping camera in place
- panorama stitching without focus stacking but including refocusing at each frame and with lots of frame overlap (so that the focus breathing is automatically corrected when the stitcher adapt focal length between frames).

[Wheatfield]

Then your best option is focus stacking - to which I'll add, a lens with a fairly flat field would be advantageous here.

The tilt lens option is an interesting possibility too, but won't be suitable for every type of scene since you'd be altering the plane of focus, not increasing the depth of field for the whole frame...

Beyond that, my congratulations - you have successfully stretched your system as far as possible for this particular application

The typical method to secure sufficient depth of field in landscape photography is to tilt the film plane forwards. It's not increasing depth of field, per se, it is bringing the image plane into the depth of field of the lens. The way to emulate this with a tilt/shift lens is to tilt the lens up and them tilt the whole rig forwards.

[biz-engineer]

I do lot's of focus stacks so that I can use the best aperture and still secure depth of field.

Problem with the K1 is the focus stacking is pretty much manual. I've done 3 x image focus stack (foreground, middle-ground and background) but the result was not good. Three image stack is not enough when there is a lot of close foreground, I'd probably need a dozen of images stack to have smooth focus stacking result.

[Wheatfield]

Aperture bracketing would be a partial approach. If use my 100 macro lens for wide angle panorama, I need to stop down the lens to f16 (for example) due to the 100mm focal lens DoF is reduced, so I need to stop down the lens as if I was shooting with a 4x5 camera. With f16, my 5um K1 pixels are blurred by diffraction, so f16 improves my DoF but destroys my resolution. If I use f8, the center of the image (focus with center point) is in focus and not so much affected by diffraction, but the outer part of the image is out of focus. If I blend the f16 image with the f8 image, I can keep the center of the frame from the f8 image and the replace the out of focus part of the image by the f16 frame parts. Doing so, I avoid diffraction in the most in focus image aera, and I let more diffraction be in regions where focus is not as good. This avoid to have diffraction hit 100% of the frame.

Practical example: I have a big rock in the foreground with lots of pixels to define that rock, and I have small trees far away in the distance with few pixels to define tree branches. I take one shot at f16 to get the rock in focus, the tree in the distance looks fuzzy due to pixel level diffraction. I take an other shot at f8, the rock in the foreground is OOF, but the tree detail is sharp. I blend the two images to keep the rock in the foreground from the f16 exposure, and keep the tree in the distance from the f8 exposure. This is not focus stacking since changing the lens aperture doesn't change the focal length of the lens, refocusing the lens in case of focus stacking would due to focus breathing.

That is a cool idea.
Have you tried Photoshop for stacking? I just take the layers, often 30 or more, and let Photoshop align them and then blend them. The software seems to do a pretty good job with the resizing , sometimes I get soft spots for no apparent reason. The next time it happens, I think I will process the blended images with the in focus images that got omitted and see what happens.

---------- Post added Apr 24th, 2020 at 12:00 AM ----------

That aperture bracketing thing does actually work. I have used it few times in the same manner - to get "sharp" areas around the critically sharp primary subject in a landscape scene. It is useful in another technique as well - sometimes trying to focus stack with a telephoto lens the foreground may have parts which cover other distant parts of the image. Then, even the best focus stacking software will have problems trying to blend in the foreground with the background when large aperture is used. But if one stops down enough for the foreground parts, it will ease the blending job for the stacking software.

--

An example of aperture bracketing with focus stacking:



Here I used 100mm WR Macro to take a good bunch of f/5.6 exposures at different focusing distances closing the aperture down towards f/11 when focusing got closer to camera. There are no foreground parts covering background areas but stopping down more towards close distance radically lowered the amount of exposures needed. Helicon Focus was used to stack the exposures.

Wow. That's gorgeous.

[Wheatfield]

Problem with the K1 is the focus stacking is pretty much manual. I've done 3 x image focus stack (foreground, middle-ground and background) but the result was not good. Three image stack is not enough when there is a lot of close foreground, I'd probably need a dozen of images stack to have smooth focus stacking result.

Having to refocus the lens by hand requires a careful touch and a solid tripod. This is a 28 image stack staken with the K1 and DFA* 50/1.4 Most of the images were from close to the camera, I like your aperture stacking idea. I wouldn't mind cutting the number of exposures I have to work with a little bit. I didn't quite get the top left hand corner, but I like this crop. There is also a little bit of blur in the tips of the grass.

[biz-engineer]

Having to refocus the lens by hand requires a careful touch and a solid tripod. This is a 28 image stack staken with the K1 and DFA* 50/1.4

28 images stack is a lot of image captured manually. It would be nice if we could defined the start and stop of focus position and have the camera perform the steps in between. Aperture bracketing works, using my phone as a mean of remote control I can layer frames with simple gradient mask, without needing frame alignment. Sometimes the connection between my phone (image sync) and the camera hangs so I have to reset it all. I would mind if image sync wouldn't download image previews from the camera SD cards to the phone when I don't ask for it. The PC application (Wifi commander for Pentax, written by AndreaK) works a lot better than image sync, and also has a mode for automation, which is great, but I'd have to carry my laptop along the camera in the field.

[Wheatfield]

28 images stack is a lot of image captured manually. It would be nice if we could defined the start and stop of focus position and have the camera perform the steps in between.

Oh yeah. Doesn't Olympus do that and with their super resolution thing?

[biz-engineer]

Doesn't Olympus do that and with their super resolution thing?

No sure about Olympus. I think Fuji and Nikon are going it. Nikon's setup is a little weird (from the settings you don't really know how many shots it will focus shift and capture), it does it to infinity, so you may end-up with 100 shots or more when you need only 30.

---------- Post added 24-04-20 at 08:42 ----------

I use PS a lot when exposing for focus stacks. Helicon Focus seems to be able to take advantage of added details depending on stacking mode. Method B and C produce less errors.

What nobody says in the camera industry is how bayer interpolation is the reason why you should print at resolution between 200 and 300 PPI, while you only need 100 to 150 PPI when the camera sensor would have tri-color pixels. Example: any picture from any bayer camera looks good when zoomed at 50% on a monitor, the monitor itself look sharp when displaying of a synthesized image at 100%, but any camera image doesn't look as sharp on display unless 2 camera pixels are used to make 1 display pixel. When camera makers sell you a 36Mpixels camera, it's actually a 9Mpixels RGB camera. People say "you only need 8 mega pixels for prints"... that's true if you consider that a 4K monitor has 8 mega of three color RGB dots (so it's actually much more than 4K single dots...). A 4K monitor actually resolves as much as a 24Mpixel bayer sensor camera.