Hi, nice experiment. Since you are doing flat-stitching with the shift lens, you should use Reposition layout in the Photo Merge dialog. This is the only option that doesn't distort the images, it only aligns them using move + rotate operations. Other options are for rotational panoramas when you rotate the camera around a nodal (non-parallax) point on a panoramic head. And I am afraid you can't get true 1 gigapixel image because you just don't have enough pixels with shift stitching - the pixels you got from perspective merge are not true, they are interpolated. If you shift the lens by +/- 12 mm, you basically get virtual sensor area of 44 × (33 + 2 × 12) mm = 44 × 57 mm, which is 1.727 times larger area than the GFX sensor. So you "only" have 400 × 1.727 = 690 megapixels of real (captured) pixels.
Although I am a high school sports/senior portrait photographer, I really enjoy all your videos and your in-depth explanations. Also, I have visited Switzerland many years ago when I lived in Germany and enjoy seeing the quaint villages against the snow capped peaks. Here in central east coast of Florida, it is quite different but I love our scenery as well. Thanks so much, Ted in Sebastian, Fl.
Thanks for the nice video, interesting to see your process. Personally, I find using pixel shift a bit too delicate and uncertain, as it works only with a super static camera and subject and you can't review the result straight away in camera. My own process is different. I picked up a second hand Syrp Genie II Pan-Tilt and use PTGui to have maximum control over photo stitching. This works amazing, especially with telelenses! I started doing it with the Sony A7Riii combined with the 135mm f1.8 GM, and am currently using the GFX100s with the GF 250mm f4 as well. I just shoot the regular resolution of the sensor (no pixel shift) and PTGui instantly works like a charm for stitching dozens of photos together as long as they have some reference points in them (blue skies etc. is an issue). The stitching you can perform with RAWs and you can export a single 16 bit TIFF, keeping all information in there before doing your final edit in Photoshop or Lightroom (where you're limited to 512 Megapixel indeed). I have had great results with this, and the process is quick, easy and more certain (and forgiving, as you can swap a single photo out from a panorama).
Thank you for another inspiring video. Your panorama images are sharp as a tack. A number of years ago I got the panorama bug after viewing sample panorama images I believe were taken, if I recall, with a Canon 5d using a telephoto lens. The images were processed using a software called Autopano Pro and were simply amazing. So I took the dive and bought the software. For me the only downside, if you want to call it that, was financial. I soon discovered that I didn’t have enough processing power or patience to wait for my Mac Book to complete the monumental task of stitching at least 20 or more images together. So for the sake of experimentation and what turned out to be a totally enjoyable but short lived phase in my photographic history, I purchased a Mac Pro with an enormous number of cores. I was in heaven. I still enjoy creating panoramas on a smaller scale with my Leica M11 and my nephew is now enjoying the Mac Pro for his musical endeavors. Keep the great videos coming.
Thanks William, much appreciated comment. I did the processing on my Macbook Pro 16" with M1 Max and 64GB RAM, works without flaws. But your comment is spot-on, you need a lot of computational power to process such pixel dimensions. Your Mac Pro for sure will do even better than my Macbook :)
Great video. Many thanks for sharing. May I ask which tripod and head combo you used? You have excellent sharpness in your images. Also, did you process the night time images in Capture One or Lightroom/Photoshop?
Interesting thanks. I'm always wondering what the best way to process my stitched panos is and avoiding parallax error. A shift lens sounds fun to play with.
You can basically do this with any camera, but try to use a telephoto lens up to 85 - 200mm Full Frame equivalent (whatever you have available). The 50mm in the video is quite short (but not for the pixel shift technique used). Check the horizontal view angle of the lens and and shoot with about 20-30% overlap between photos. E.g. if the angle of view is 20 degrees, take a photo every 15 degrees on the pan scale of your tripod head. Use quite a large depth of field, f5.6 - f11 and fix your ISO and white balance to 1 setting. This allows for photo software to stitch the pics quite easily, and because you stitch close-ups the details will be amazing. You can likely try this with your current camera gear and any of the mainstream photo software, without having to make any investments, and get some great results. I have used a 24 megapixel APS-C camera for this in the past too, works just fine and the results are fun!
Hi! Would you recommend using a Canon TS lens on an SL2? Is there an adapter? Would that even make sense? Thanks for the video! Interesting, as always!
Makes perfectly sense to use Canon TS lenses on the SL2, Tom. There are also various adapters you can use for it. Canon TS lenses have a very large image circle, bigger than what you need for full frame sensors. On the SL2 you, therefore, get the best out of these lenses.
You mentioned in the video that there was no vignetting. Can you use the full shifting distance of I believe 12 mm each way without vignetting? That would be awesome, given the full frame lens on medium format sensor.
I set it to full medium format sensor use, no crop setting. You see it best at the end of the video when I show the single frame of the monastery without shifting: the aspect ratio is 4:3 means medium format and no crop.
Do you think a digital cameras from phones will be ever close to a high-quality photography? There are news about a 200 mpix cameras incoming. I know I am not saying much but you are more an expert, than me overall.. so you know more about what is coming with what specs than me.
The short answer is "no". With phones you will always be restricted by the size of the phone and the budget they can allocate to the camera hardware and software. It will always be a small sensor or sensors, and small thin lenses, which just give physical limitations on what can be achieved. While phones get ever better in mimicking with software what we find desirable in images taken with real cameras and good glass, I don't think they will ever become a full substitute for high quality photo gear. But I do think it puts something very powerful in the hands of laymen, while working with photo gear takes time to learn. For the low(er) quality photo gear - the traditional holiday point-and-shoot - I think the phone has for the most part already overtaken it and those cameras are on their way out as most people may get better results with a phone than with a low end camera they rarely use. I know on CZcams they like to find some test where the phone beats a fancy camera. But you buy a fancy camera to get good results in almost all situations, while with a phone such scenarios are much more limited. It is all about the right tool for the job.
No - smartphones in the next years will not be able to replace pro cameras. Smartphones can do a lot these days via computational photography, and its fascinating how these little computers take photos. But the reality of physics and optics remains so that smartphones will not make pro cameras redundant.
Good work! You can check out Microsoft ICE (discontinued MS Research project) but still can be found is a great tool to go over 1GPx, it's really quick end efficient program. There is also Hugin - free, but I not really like it or PTGui - paid software, never used it.
Hi, nice experiment. Since you are doing flat-stitching with the shift lens, you should use Reposition layout in the Photo Merge dialog. This is the only option that doesn't distort the images, it only aligns them using move + rotate operations. Other options are for rotational panoramas when you rotate the camera around a nodal (non-parallax) point on a panoramic head. And I am afraid you can't get true 1 gigapixel image because you just don't have enough pixels with shift stitching - the pixels you got from perspective merge are not true, they are interpolated. If you shift the lens by +/- 12 mm, you basically get virtual sensor area of 44 × (33 + 2 × 12) mm = 44 × 57 mm, which is 1.727 times larger area than the GFX sensor. So you "only" have 400 × 1.727 = 690 megapixels of real (captured) pixels.
Excellent information as always! I always look forward to your videos. I enjoyed this GigaPixel challenge.
Although I am a high school sports/senior portrait photographer, I really enjoy all your videos and your in-depth explanations. Also, I have visited Switzerland many years ago when I lived in Germany and enjoy seeing the quaint villages against the snow capped peaks. Here in central east coast of Florida, it is quite different but I love our scenery as well. Thanks so much, Ted in Sebastian, Fl.
Thank you Ted, appreciate your kind comment and feedback.
Thanks for the nice video, interesting to see your process. Personally, I find using pixel shift a bit too delicate and uncertain, as it works only with a super static camera and subject and you can't review the result straight away in camera. My own process is different. I picked up a second hand Syrp Genie II Pan-Tilt and use PTGui to have maximum control over photo stitching. This works amazing, especially with telelenses! I started doing it with the Sony A7Riii combined with the 135mm f1.8 GM, and am currently using the GFX100s with the GF 250mm f4 as well. I just shoot the regular resolution of the sensor (no pixel shift) and PTGui instantly works like a charm for stitching dozens of photos together as long as they have some reference points in them (blue skies etc. is an issue). The stitching you can perform with RAWs and you can export a single 16 bit TIFF, keeping all information in there before doing your final edit in Photoshop or Lightroom (where you're limited to 512 Megapixel indeed). I have had great results with this, and the process is quick, easy and more certain (and forgiving, as you can swap a single photo out from a panorama).
Great comment and hints, many thanks, Edwin.
Excellent video
Beautiful images !!!
Thanks for sharing.
Thanks RS!
Thank you for another inspiring video. Your panorama images are sharp as a tack. A number of years ago I got the panorama bug after viewing sample panorama images I believe were taken, if I recall, with a Canon 5d using a telephoto lens. The images were processed using a software called Autopano Pro and were simply amazing. So I took the dive and bought the software. For me the only downside, if you want to call it that, was financial. I soon discovered that I didn’t have enough processing power or patience to wait for my Mac Book to complete the monumental task of stitching at least 20 or more images together. So for the sake of experimentation and what turned out to be a totally enjoyable but short lived phase in my photographic history, I purchased a Mac Pro with an enormous number of cores. I was in heaven. I still enjoy creating panoramas on a smaller scale with my Leica M11 and my nephew is now enjoying the Mac Pro for his musical endeavors. Keep the great videos coming.
Thanks William, much appreciated comment. I did the processing on my Macbook Pro 16" with M1 Max and 64GB RAM, works without flaws. But your comment is spot-on, you need a lot of computational power to process such pixel dimensions. Your Mac Pro for sure will do even better than my Macbook :)
Truely extraordinary
Many thanks!
Great video. Many thanks for sharing. May I ask which tripod and head combo you used? You have excellent sharpness in your images. Also, did you process the night time images in Capture One or Lightroom/Photoshop?
Interesting thanks. I'm always wondering what the best way to process my stitched panos is and avoiding parallax error. A shift lens sounds fun to play with.
Shift lenses are real fun :) Best wishes for YOUR Gigapixel journey :)
Fascinating, love the video
Great - thx!
It's preferable to fix the lens to the tripod and shift the camera to avoid parallax issues. You need a lens frame for this.
Very interesting!
How do you compare Image quality between fujifilm GF45 and canon 50mmTS-E ?
What is your setting with the gfx, ibis on/of, diafragma minimum and other tips? Can’t get it tag sharp with my GFX100S viltrox and 24 test II
IBIS on tripod needs to be set to off. Only when shooting hand-hold IBIS should be on.
Beautiful! The level of detail is insane! 😍 I don't want to know how much would cost this combination of camera, adapter and lens...
You can basically do this with any camera, but try to use a telephoto lens up to 85 - 200mm Full Frame equivalent (whatever you have available). The 50mm in the video is quite short (but not for the pixel shift technique used). Check the horizontal view angle of the lens and and shoot with about 20-30% overlap between photos. E.g. if the angle of view is 20 degrees, take a photo every 15 degrees on the pan scale of your tripod head. Use quite a large depth of field, f5.6 - f11 and fix your ISO and white balance to 1 setting. This allows for photo software to stitch the pics quite easily, and because you stitch close-ups the details will be amazing. You can likely try this with your current camera gear and any of the mainstream photo software, without having to make any investments, and get some great results. I have used a 24 megapixel APS-C camera for this in the past too, works just fine and the results are fun!
@@Edwin-- Agree - you can do this with much cheaper equipment than I used ...
Thanks! Edwin's comment to your comment is a good comment, you not necessarily need expensive equipment to go for huge panos
Genius
Hi! Would you recommend using a Canon TS lens on an SL2? Is there an adapter? Would that even make sense? Thanks for the video! Interesting, as always!
You can use the Canon TS on SL2 with the Sigma MC21 on any L mount body.
Makes perfectly sense to use Canon TS lenses on the SL2, Tom. There are also various adapters you can use for it. Canon TS lenses have a very large image circle, bigger than what you need for full frame sensors. On the SL2 you, therefore, get the best out of these lenses.
Try Hugin and see if they can process these large files.
You mentioned in the video that there was no vignetting. Can you use the full shifting distance of I believe 12 mm each way without vignetting? That would be awesome, given the full frame lens on medium format sensor.
I stopped at 10mm but until then I had absolutely no vignetting.
@@mathphotographer thanks, that sounds promising!
how about some samples?
Did you shoot with 35mm, Off, or Auto in menu setting ?
I set it to full medium format sensor use, no crop setting. You see it best at the end of the video when I show the single frame of the monastery without shifting: the aspect ratio is 4:3 means medium format and no crop.
@@mathphotographer Thanks. End result is superb.
Do you think a digital cameras from phones will be ever close to a high-quality photography?
There are news about a 200 mpix cameras incoming. I know I am not saying much but you are more an expert, than me overall.. so you know more about what is coming with what specs than me.
The short answer is "no". With phones you will always be restricted by the size of the phone and the budget they can allocate to the camera hardware and software. It will always be a small sensor or sensors, and small thin lenses, which just give physical limitations on what can be achieved. While phones get ever better in mimicking with software what we find desirable in images taken with real cameras and good glass, I don't think they will ever become a full substitute for high quality photo gear. But I do think it puts something very powerful in the hands of laymen, while working with photo gear takes time to learn.
For the low(er) quality photo gear - the traditional holiday point-and-shoot - I think the phone has for the most part already overtaken it and those cameras are on their way out as most people may get better results with a phone than with a low end camera they rarely use.
I know on CZcams they like to find some test where the phone beats a fancy camera. But you buy a fancy camera to get good results in almost all situations, while with a phone such scenarios are much more limited. It is all about the right tool for the job.
No - smartphones in the next years will not be able to replace pro cameras. Smartphones can do a lot these days via computational photography, and its fascinating how these little computers take photos. But the reality of physics and optics remains so that smartphones will not make pro cameras redundant.
@@Edwin-- & @mathphotographer Thank you for that long answer
Have you tried 16 bit HDR files with gigapixel shots?
Not yet - but clearly coming with even more computational effort
I USE STILL AUTOPANO GIGA BEST FOR STICHING
Good work! You can check out Microsoft ICE (discontinued MS Research project) but still can be found is a great tool to go over 1GPx, it's really quick end efficient program. There is also Hugin - free, but I not really like it or PTGui - paid software, never used it.
Thanks for pointing out, will look into it.