Yeah, I recently changed my setting to do just what you mentioned - turn various manual control dials right to bring the exposure meter right. Anything that makes it quicker to adjust consistently is appreciated as I get older!
Upvote
0
It’s time to fill those memory cards. Canon releases firmware v1.8.1 for the Canon EOS R5. 400mp stills are now possible
- Thread starter Canon Rumors Guy
- Start date
Flash is not essential for macro; and in camera focus bracketing is a relatively recent feature, no reason you can't do it the old fashioned way (although good luck to anyone making a focus stack with 400MP sub frames).
Upscaling doesn't generate extra detail, just the illusion of it, so while that may be useful for many people and purposes, it's not a substitute for others.
Upvote
0
Perhaps they will add the ability to use flash in a future iteration, do any of the other brands with super-resolution imaging allow it?
Upvote
0
I’m pretty certain it won’t ever happen for the R5. The limitation is that the R5 cannot use flash with electronic shutter, and pixel shift requires electronic shutter to avoid the vibration that would be introduced from the mechanical shutter opening and closing.
If they add pixel shift to the R3 (which does allow flash with electronic shutter), we will see if flash is possible with pixel shift.
Upvote
0
I thought about test images and the new 400M jpg option, and decided - "What the 'ell, let's do this!".
So I dusted off my RF 85 1.2L as my "best testable" lens, and put my R5 on a good (but not concrete) tripod with *no* legs extended, on my basement concrete floor with the garage door open to illuminate everything with just diffuse sunlight (*no* electric light).
I took my previously designed Lens Test Chart and leaned it against 2 plastic storage containers, and lightly held a laser emitting tape measure perpendicular to the test chart with the beam into the center of the lens (as much as possible).
Then I set the R5 to f2.8 and took the best 45M shot of it with a wired remote. This is the (un-cropped) photo of the test chart, after being output by my DXO Photolab Deep Prime output. The 8192 x 5464 size is reduced to 2048 to fit in the size allowed for uploading:
I cropped the above image to just show the detailed test chart. If you want some fun, you can "solve" the equations by reading the instructions, and find out how many (black & white) lines can be resolved (enough to recognize for sure) by my setup. The following is this crop, also size reduced so that the height is 2048 to fit in the upload:
Next, I further cropped the image to just show 1 square of test lines I created & printed on my laser printer. This is now a 1-to-1 (NOT resized) image of just it. If you look into it, you will see I can resolve 2 lines (1 white & 1 black) within each 3!!! pixels. You can tell the lines are there, within 3 pixels, even though it's not pure black & white at that point. The resolution accuracy is pretty much the same spread across the entire 8K original image that I checked. Now, you can use this image, along with the instructions, and see how many lines (1 white & 1 black = 2 lines) of resolution you have from the top of the sensor to the bottom (5464 pixel high sensor):
(Note: I seem to have cropped off the image for the "A" line - sorry. But the previous image can be used to do calculations, too!)
In addition, I did a similar 1-to-1 crop of the queen's portrait, which you can see here:
It was fun to try this, to see what my R5 setup can do.
I didn't show the equations being solved here, so you can do that yourself if you want some fun.
The next thing to do would be to do the same with a 400M image. That will be the next thing I get to (well, maybe today, but who knows?). I wonder what it will look like?
Cheers!
So I dusted off my RF 85 1.2L as my "best testable" lens, and put my R5 on a good (but not concrete) tripod with *no* legs extended, on my basement concrete floor with the garage door open to illuminate everything with just diffuse sunlight (*no* electric light).
I took my previously designed Lens Test Chart and leaned it against 2 plastic storage containers, and lightly held a laser emitting tape measure perpendicular to the test chart with the beam into the center of the lens (as much as possible).
Then I set the R5 to f2.8 and took the best 45M shot of it with a wired remote. This is the (un-cropped) photo of the test chart, after being output by my DXO Photolab Deep Prime output. The 8192 x 5464 size is reduced to 2048 to fit in the size allowed for uploading:
I cropped the above image to just show the detailed test chart. If you want some fun, you can "solve" the equations by reading the instructions, and find out how many (black & white) lines can be resolved (enough to recognize for sure) by my setup. The following is this crop, also size reduced so that the height is 2048 to fit in the upload:
Next, I further cropped the image to just show 1 square of test lines I created & printed on my laser printer. This is now a 1-to-1 (NOT resized) image of just it. If you look into it, you will see I can resolve 2 lines (1 white & 1 black) within each 3!!! pixels. You can tell the lines are there, within 3 pixels, even though it's not pure black & white at that point. The resolution accuracy is pretty much the same spread across the entire 8K original image that I checked. Now, you can use this image, along with the instructions, and see how many lines (1 white & 1 black = 2 lines) of resolution you have from the top of the sensor to the bottom (5464 pixel high sensor):
(Note: I seem to have cropped off the image for the "A" line - sorry. But the previous image can be used to do calculations, too!)
In addition, I did a similar 1-to-1 crop of the queen's portrait, which you can see here:
It was fun to try this, to see what my R5 setup can do.
I didn't show the equations being solved here, so you can do that yourself if you want some fun.
The next thing to do would be to do the same with a 400M image. That will be the next thing I get to (well, maybe today, but who knows?). I wonder what it will look like?
Cheers!
Last edited:
Upvote
0
OK - I went and took a 400MP pixel-shift image (well, actually I did it two different times, and they seemed to have a bit of a color shift, probably due to clouds coming & going as I use only outdoor lighting coming in the garage opening (no electric lights to destroy the pixel-shifted shots).
I previously posted 4 images for my R5, RF 85mm f1.2L at f2.8 normal 45MP (23MB image).
Now I have taken two 400M images of the same thing (nothing moved!). Since they are 3x the size & width, the closeup crops of the lines and the Queen were much larger than 2K, so to keep 1:1 pixels I had to crop a smaller version of both so that the size of a square for them was ~2K x 2K pixels. I also wanted to show the 45M version crop output (from DXO PhotoLab) at 3x bigger dimensions to match the size from the 400M crop so I had to crop that image smaller to match the framing of the 400M crops. I also tweaked the white balance a smudge to get the 3 versions to look a little more similar to each other (I might have gotten a couple a little off there).
The resultant 400MP files were 24576 × 16384, and 105MB. There's no point to showing resized-down 2K versions of the 400M entire images, so I won't show you those.
For the 1:1 (not resized) lens test lines crop, I'll first show the 2 new line crops of the 400M image:
As a comparison, I cropped the same lines image from the 45M file and interpolated them up by 3x to match the image and #pixels from the 400M file. So now they can be compared against each other: Here's the 45M 3X upres'd version:
Now, I'll show the Queen crop from the two 400M images. They're 1:1 (not resized):
Now I'll show the 3X upres'd version of the 45M image:
OK - THE RESULT ...
I think the 400M gives a better photo, hands down! The file storage size is 4x as large for me. If I wanted a single shot of the Grand Canyon on a great tripod & calm (non-windy) day, then YES, I am surprising myself and I would use this feature. But it won't be for anything that moves, which is really most of the things I do.
There is more resolution in the photo, but (of course) not what I would expect with a true (future) Canon 400M sensor. But it really is better than the 45MP shot, so I will give Canon more praise than jeers on this one. I didn't get any of the crazy artifacts that some others have seen (don't use electric lighting source, maybe).
So, what do you think?
I previously posted 4 images for my R5, RF 85mm f1.2L at f2.8 normal 45MP (23MB image).
Now I have taken two 400M images of the same thing (nothing moved!). Since they are 3x the size & width, the closeup crops of the lines and the Queen were much larger than 2K, so to keep 1:1 pixels I had to crop a smaller version of both so that the size of a square for them was ~2K x 2K pixels. I also wanted to show the 45M version crop output (from DXO PhotoLab) at 3x bigger dimensions to match the size from the 400M crop so I had to crop that image smaller to match the framing of the 400M crops. I also tweaked the white balance a smudge to get the 3 versions to look a little more similar to each other (I might have gotten a couple a little off there).
The resultant 400MP files were 24576 × 16384, and 105MB. There's no point to showing resized-down 2K versions of the 400M entire images, so I won't show you those.
For the 1:1 (not resized) lens test lines crop, I'll first show the 2 new line crops of the 400M image:
As a comparison, I cropped the same lines image from the 45M file and interpolated them up by 3x to match the image and #pixels from the 400M file. So now they can be compared against each other: Here's the 45M 3X upres'd version:
Now, I'll show the Queen crop from the two 400M images. They're 1:1 (not resized):
Now I'll show the 3X upres'd version of the 45M image:
OK - THE RESULT ...
I think the 400M gives a better photo, hands down! The file storage size is 4x as large for me. If I wanted a single shot of the Grand Canyon on a great tripod & calm (non-windy) day, then YES, I am surprising myself and I would use this feature. But it won't be for anything that moves, which is really most of the things I do.
There is more resolution in the photo, but (of course) not what I would expect with a true (future) Canon 400M sensor. But it really is better than the 45MP shot, so I will give Canon more praise than jeers on this one. I didn't get any of the crazy artifacts that some others have seen (don't use electric lighting source, maybe).
So, what do you think?
Last edited:
Upvote
0
The same results with the same technique? Yes! Well, what I did wasn't rocket science. But the test pattern did take me a long time.
But anyone can glue a colorful national banknote (like a British 5 Pound bill) on flat board. Get a laser distance measure to get it perpendicular with the lens (I wasn't all that precise, and it turned out pretty good for me). I just chose f2.8 and a short distance to the board. If I moved the board twice as far away, then the test pixels would be 2x smaller and allow further testing of the high resolution I'm seeing. Other f#s would have slightly different results, but I just tried my best guess where this lens would have the sharpest resolution.
I happen to have the 85mm f1.2L, which 'might' be the sharpest (or maybe one of them) Canon RF lenses, so at least it's not the limiting factor (I think). Other non-L lenses might not have the resolution to benefit as much, so you'd just have to try them and see. L zooms would work fine as long as you don't jar the lens focus travel in any way (it's just one more thing to go wrong when things are this precise). Non-L zooms at low prices might not make you happy - I don't know.
Upvote
0
I think that the resolution chart needs an update.
The only extra detail not resolved with the non-shifted sensor (and resolved with the shifted one) seems to be the "pounds" text under the ear.
Upvote
0
This is a 2 euro coin with a 100mm EF L macro lense at closest focus distance, without crop.
This is a 100% Crop on the E in the "EURO" :-D
And this is the corner of the coin:
Its crazy :-D
And money is disgusting, if you look to close at it :-D
Setup on daylight, camera and coin just put on the table. F4, ISO 640, 1/50s
This is a 100% Crop on the E in the "EURO" :-D
And this is the corner of the coin:
Its crazy :-D
And money is disgusting, if you look to close at it :-D
Setup on daylight, camera and coin just put on the table. F4, ISO 640, 1/50s
Upvote
0
Upvote
0
The resolution chart is not perfect, I grant you. It was the best that my laser printer could do. But if I move the test chart 2 times further away, you will be able to use a test chart lines # which is 2x bigger, and so the resulting line image to the sensor will have 2x or more apparent detail (to the sensor) since there was a vastly larger # of pixels to make a clean black, white & edge detail. And the same is true if you get 4x further and instead of the 80 # you look at a 20#. If you happen to know of a perfect test chart, with the equations & resolution #s for variously spaced clean line patches shown from super tiny to large resolutions, then I'd be glad to get one of them.
However, in the Queen's image I was able to see a lot more detail in the tiny/subtle parts of it. And that's at the current distance of test chart. So if (ok, when) I double (or quadruple) the test chart distance, it will be interesting to see how things look when the test chart itself will no longer be (or part of) the weak link.
Last edited:
Upvote
0
I went out to my studio this morning where I have a heavy duty tripod head bolted to a heavy light table and checked out the 400mp setting on my R5. I have bright lighting, so I could try different apertures on my 24-70 zoom and manually focus and shoot remotely while viewing at 15X.
I could see camera shake if I even moved and it took about 6 seconds to stop. By using that careful method, I managed to see that there were no artifacts, and that the lens was the limiting factor as far as acuity. I may try with my 100L, it may be a little better. This what I see with a crop viewing at 100%. It looks just like the live view screen on my pc. Dealing with the large file is just ok. Unfortunately, I don't think a distant bird will sit still enough to get all setup, but by sandbagging the tripod or hanging weight from the under hook, it may be possible to crop a distant still object and get more detail. My lens is going to be the limiting factor. I did not bother going thru the process of taking the same photo without the 400mp setting and up resolving it, I was just making sure that the image had no artifacts. It is slightly soft, that's due to the limitation of my lens at 70mm.
I could see camera shake if I even moved and it took about 6 seconds to stop. By using that careful method, I managed to see that there were no artifacts, and that the lens was the limiting factor as far as acuity. I may try with my 100L, it may be a little better. This what I see with a crop viewing at 100%. It looks just like the live view screen on my pc. Dealing with the large file is just ok. Unfortunately, I don't think a distant bird will sit still enough to get all setup, but by sandbagging the tripod or hanging weight from the under hook, it may be possible to crop a distant still object and get more detail. My lens is going to be the limiting factor. I did not bother going thru the process of taking the same photo without the 400mp setting and up resolving it, I was just making sure that the image had no artifacts. It is slightly soft, that's due to the limitation of my lens at 70mm.
Upvote
0
I tried the same with my 100L, its a bit sharper. Its also a longer focal length so its difficult to compare.
Upvote
0
My Old 50mm f/2.5 Macro produces about the same image. I moved it a bit closer on my focus rack
Upvote
0
