Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: eronald on February 15, 2018, 11:30:00 am
-
It looks like those millions of dollars of R&D invested by phone makers are starting to pay off ...
https://www.dpreview.com/news/2999078463/noa-n7-smartphone-captures-80mp-images-with-high-resolution-mode
Edmund
-
I remember when someone released a 41 (?) megapixel phone a couple years back; I downloaded a few JPGs. They were unambiguously atrocious even by phone standards.
It's the way of consumer companies to play the spec game rather than focus on actual quality; consumers don't understand "MTF" or "noise pattern characteristics" or "color filter array quality" but they do (vaguely) understand megapixels. 99.99% of consumers would be far better off with a 20 megapixel phone that had great image quality characteristics (especially low noise in low light) and usability features (e.g. faster and better on-sensor focusing) than an 80mp phone with even small compromises in those other areas – and if the 41 megapixel phone was any guide the compromises won't be small.
But I think you know that and are just trying to stir the pot. It does kind of stir me up, but not because of the odd (and entirely not serious) insult to MF. It stirs me up because it almost feels like taking advantage of consumer nativity; your grandma (or sibling, or friend) could buy such a phone thinking they are getting a better camera when in all likelihood they are getting a much worse camera than alternative options with lower megapixels, and the phone maker knows that and still chooses to make it.
-
I remember when someone released a 41 (?) megapixel phone a couple years back; I downloaded a few JPGs. They were unambiguously atrocious even by phone standards.
It's the way of consumer companies to play the spec game rather than focus on actual quality; consumers don't understand "MTF" or "noise pattern characteristics" or "color filter array quality" but they do (vaguely) understand megapixels. 99.99% of consumers would be far better off with a 20 megapixel phone that had great image quality characteristics (especially low noise in low light) and usability features (e.g. faster and better on-sensor focusing) than an 80mp phone with even small compromises in those other areas – and if the 41 megapixel phone was any guide the compromises won't be small.
But I think you know that and are just trying to stir the pot. It does kind of stir me up, but not because of the odd (and entirely not serious) insult to MF. It stirs me up because it almost feels like taking advantage of consumer nativity; your grandma (or sibling, or friend) could buy such a phone thinking they are getting a better camera when in all likelihood they are getting a much worse camera than alternative options with lower megapixels, and the phone maker knows that and still chooses to make it.
Doug,
For once you are accusing me - and I agree it is a rare thing indeed - of a provocation when I am not making one.
Let us remove our grandma from the picture - we are all pros or very serious amateurs here, and even I have some tech ability - my math degree was actually earned, and I have taken the Imatest course in Boulder so I probably know as well as most end-users what an MP number means and what MTF is. Many here are more savvy than me in tech, and most forum members here are accomplished as photographers and have some experience of medium format.
What I was trying to say is that there is a move to computational photography, multiple image compositing, with DR extension and super-resolution. In a way this is as revolutionary as going from film to digital, because intelligence and multi-sensor fusion is used to assemble an image rather than brute force - a 40MP image is assembled from eg 10 or 20 or 100 low-rez shots. The Google Pixel with its HDR, and the iPhone X with its simulated portrait lightings are harbringers of this wave.
To me, this transition to smarter imaging -MP numbers which are NOT sensor MP numbers- is a real revolution, driven by the fact that phone computers are in fact much more powerful and expensive than most camera computers. As a result, most digital backs rely on brute force sensors for acquiring imagery which is written to file with the least possible processing, while phones try and get the most work done by processing while relying on a cheap sensor.
Now my question is when are the "pro" cameras going to catch up? When will pro cameras embody computational photography? Given the computer power required - a multiple of what is necessary in a phone- I suspect the first embodiments will come in the form of tethered software or reprocessing of files from high end cine equipment.
Edmund
-
Click Bait
-
Click Bait
This phone, maybe. The tech, no. Google Pixel 2 has a custom VLSI just for imaging.
It's all about maximizing processing during capture.
Look at the sky and pole here:
https://www.popsci.com/google-pixel-visual-core-smartphone-chip#page-3
-
Hi Edmund,
It seems to me that camera sensor seem to be pretty stuck at 24MP for APS-C, 24x36 mm at 42-50MP. MF does significant progress with 100MP 44x33 and 150 54x40 on Sony's published roadmap, those MF sensors would correspond to 60MP on 24x36.
It is nice to see development on cell phones. Let us not forget that most images are actually shot on cell phones and all smart things they are putting in those cell phones actually improve all those Tera pictures (to avoid conflict between US and European numbers).
Although extreme resolution is sometimes needed, like in reproduction of artwork or arial photography, the main reason to use "real cameras" are lenses.
On the other hand, compact cameras are nearly a dead end, although high end compacts seem to sell reasonably well, I guess.
An other way to see it is that a "camera is simply an imaging device". Nothing says that you cannot attach a long lens to a cell phone.
I guess that we will see a lot of change in the camera industry with the arrival of 8K. Camera electronics that can deliver 8K at 50p will need very advanced electronics and that will be able to handle also stills.
One interesting thing is that multi frame technologies probably will play a larger role in the future. 8K needs like 39MP on 3:2 formats. You shoot at 50 frames/s, that is a lot of data, to used with supersampling.
It obviously doesn't work with studio flash, but I guess that continuous light will see more and more use because of video.
I have read that Forza Semiconductor has developed a 133 MP sensor for NHK. It will be used for broadcasting. It seems that it has 43 mm diagonal but will obviously use 16/9 format rather than 24x36. It samples an RGBG quartet for each 8K pixel. DR will of course be a bit limited compared with larger pixels.
That sensor would have around 2.5 micron pixels, pretty ideal for today's best lenses.
Best regards
Erik
It looks like those millions of dollars of R&D invested by phone makers are starting to pay off ...
https://www.dpreview.com/news/2999078463/noa-n7-smartphone-captures-80mp-images-with-high-resolution-mode
Edmund
-
I guess you can attach the same processing chippery to a 20MP APS-C, or full 35 chip like the A7S, as to a cellphone. In fact I would be surprised if Sony didn't do it.
Edmund
Hi Edmund,
It seems to me that camera sensor seem to be pretty stuck at 24MP for APS-C, 24x36 mm at 42-50MP. MF does significant progress with 100MP 44x33 and 150 54x40 on Sony's published roadmap, those MF sensors would correspond to 60MP on 24x36.
It is nice to see development on cell phones. Let us not forget that most images are actually shot on cell phones and all smart things they are putting in those cell phones actually improve all those Tera pictures (to avoid conflict between US and European numbers).
Although extreme resolution is sometimes needed, like in reproduction of artwork or arial photography, the main reason to use "real cameras" are lenses.
On the other hand, compact cameras are nearly a dead end, although high end compacts seem to sell reasonably well, I guess.
An other way to see it is that a "camera is simply an imaging device". Nothing says that you cannot attach a long lens to a cell phone.
I guess that we will see a lot of change in the camera industry with the arrival of 8K. Camera electronics that can deliver 8K at 50p will need very advanced electronics and that will be able to handle also stills.
One interesting thing is that multi frame technologies probably will play a larger role in the future. 8K needs like 39MP on 3:2 formats. You shoot at 50 frames/s, that is a lot of data, to used with supersampling.
It obviously doesn't work with studio flash, but I guess that continuous light will see more and more use because of video.
I have read that Forza Semiconductor has developed a 133 MP sensor for NHK. It will be used for broadcasting. It seems that it has 43 mm diagonal but will obviously use 16/9 format rather than 24x36. It samples an RGBG quartet for each 8K pixel. DR will of course be a bit limited compared with larger pixels.
That sensor would have around 2.5 micron pixels, pretty ideal for today's best lenses.
Best regards
Erik
-
Hi,
Yes and no. The A9 can shoot 20FPS and the A7R2 10FPS, a major from the A7rII. They probably use the same 'front size ASIC'. But the A9 has a stacked sensor with extremely fast readout.
Some things are possible, some are not. I guess that companies like Sony plan like 3-5 years ahead and try to reuse technology when possible.
Best regards
Erik
I guess you can attach the same processing chippery to a 20MP APS-C, or full 35 chip like the A7S, as to a cellphone. In fact I would be surprised if Sony didn't do it.
Edmund
-
I agree in general that computational photography may be the future, but I’m not quite sure it’s a competitor to larger sensors (APS-C, FF 35mm, MF, etc) currently. Reviews that I’ve seen of the Light L16, for example, haven’t been great. There’s potential there but I think it needs a lot of improvement, and the computational algorithms used are far from perfect in today’s implementations. At the moment, I’ll take my X1D any day (or even stitch with APS-C) over a phone with an 80mp mode or a Light L16 when I need high resolution work.
I very much enjoy portrait mode on my iPhone X, for example, which I’m sure has to use some form of computational photography to do what it does, and it can produce great results. But even with that shooting mode it shows its limitations pretty quickly and easily, and I doubt those images that turn out well would even print that great (although maybe I should print a few 8x10s for myself to see).
OP poses the question “when will MF (or even larger sensors) catch up”, but I still think a lot of work needs to be done with computational photography before it really even becomes something to catch up to. All that said, it surely will improve in the future though and I’m looking forward to seeing it mature....
-
I submit that the feature that may sell best with the public is not the 80 MP but the "Face Beauty" processing on the fly. ;D
-
I find this topic to be quite amusing and a comical diversion from fielding emails from clients.
This week I was waiting for feedback and edit requests from two different clients after sending them 1500x2000 pixel proofs. No comments or edits had anything to do with resolution, and, to be honest, I can't remember the last time I had someone ask me what resolution I shoot at. Been at least 5 or 6 years.
I also submitted a bid on a fairly large ad job, well into 6 figures. Not even with this project was resolution a concern. We did talk about possible video and the request for 4K came up, but it was more geared towards, "can we still get good 4K without using a RED and incurring that expense?," and then moved onto color, lighting and logistics, and everything else that comes with a giant project.
The fact that people still are concerned about resolution today baffles me every time I see conversations like this.
On the ad job, one of the primary reason I was asked to bid was with how I handle color. Not once was I asked about how big my files are.
-
I find this topic to be quite amusing and a comical diversion from fielding emails from clients.
This week I was waiting for feedback and edit requests from two different clients after sending them 1500x2000 pixel proofs. No comments or edits had anything to do with resolution, and, to be honest, I can't remember the last time I had someone ask me what resolution I shoot at. Been at least 5 or 6 years.
I also submitted a bid on a fairly large ad job, well into 6 figures. Not even with this project was resolution a concern. We did talk about possible video and the request for 4K came up, but it was more geared towards, "can we still get good 4K without using a RED and incurring that expense?," and then moved onto color, lighting and logistics, and everything else that comes with a giant project.
The fact that people still are concerned about resolution today baffles me every time I see conversations like this.
On the ad job, one of the primary reason I was asked to bid was with how I handle color. Not once was I asked about how big my files are.
You're quite right Joe. But that doesn't mean that it never comes up with others. I'm currently working with 3 shooting clients right now who have had client requests for at least 80MP (and in one case, they just said as many megapixels as possible). In all 3 cases, these are commercial shooters. I guess for at least 2 of them, I have them covered with the Noa N7.
This doesn't mean that the requests are valid - as in do they really need that many megapixels, but apparently they feel strong enough to put their foot down and so it goes.
Steve Hendrix/CI
-
I remember when someone released a 41 (?) megapixel phone a couple years back; I downloaded a few JPGs. They were unambiguously atrocious even by phone standards.
There were 2 such phones, the Nokia Lumia 1020 and 808 PureView. The quality was not that bad, although not as good as MF. Interested reader can download images from flickr camera finder: https://www.flickr.com/cameras/nokia/lumia_1020/
-
What I was trying to say is that there is a move to computational photography, multiple image compositing, with DR extension and super-resolution. In a way this is as revolutionary as going from film to digital, because intelligence and multi-sensor fusion is used to assemble an image rather than brute force - a 40MP image is assembled from eg 10 or 20 or 100 low-rez shots.
(...)
Now my question is when are the "pro" cameras going to catch up? When will pro cameras embody computational photography? Given the computer power required - a multiple of what is necessary in a phone- I suspect the first embodiments will come in the form of tethered software or reprocessing of files from high end cine equipment.
Actually, MF cameras were the first to increase resolution by pixel-shift and large format had scanning backs, so the answer to your question depends on how you define "extend resolution by computational image assembly". And yes: it was done "in the form of tethered software".
Now, it may look as if I am joking, but not as much as you may think. The truth is that the camera makers do not extend resolution if there is not a market for it. The first market for large resolution was and still is MF, with a 400 mpix pixel-shift back announced for next month. Phone makers could do the same, but there is a reason why the phone you presented comes from an unknown player and not Apple or Samsung: the only reason for that higher resolution is bragging rights (and probably make a quick buck by increasing the value of the shares). The average phone user does not need it.
This does not mean that computational photography is not the future. It most probably is, but with quite different uses than increased resolution or dynamic range. Some uses have been presented in the iPhone X, maybe you have noticed?
-
It looks like those millions of dollars of R&D invested by phone makers are starting to pay off ...
https://www.dpreview.com/news/2999078463/noa-n7-smartphone-captures-80mp-images-with-high-resolution-mode
Edmund
This phone uses sensor shift to get that 80MP, apparently needing a tripod. So MF is already there (Hasselblad at least) as are Olympus, Panasonic and Pentax. The only extra I see here is using two complete cameras mounted side-by-side, and it should be clear why MF cameras are not doing that.
Perhaps you should be telling Canon and Nikon that they "need to catch up" on sensor shifting.
Update: added Panasonic
-
Even at the very highest level, i.e. international main Gucci campaigns, photographers are shooting Nikon D810 and D850.
Pixel count, up to a point, is just manufacturer marketing and sales pitch and personal taste. In the end, it's irrelevant. Don't fall for it.
If you just like the look of it, fine. But it's sort of a luxury that isn't really necessary in the majority of cases. Maybe there are some cases where it's more than a luxury but they are very few.
-
There are already full specifications:
https://www.pdevice.com/product/noa-n7-price-specs
-
Where Medium Format main players have failed is that they've piled all their money into backs and sensors and in the case of Phase One, the XF Body. But the blue rings lenses aren't massively better, they are the same optical designs as before just tarted up. Hasselblad lenses are decent but again, they are aging and aren't looking very good next to the best 35mm lenses.
IMO, it's the lenses that need the most attention. Forget about resolution for a while - 100 and incoming 150 is already mostly unnecessary for the time being.
-
Where Medium Format main players have failed is that they've piled all their money into backs and sensors and in the case of Phase One, the XF Body. But the blue rings lenses aren't massively better, they are the same optical designs as before just tarted up. Hasselblad lenses are decent but again, they are aging and aren't looking very good next to the best 35mm lenses.
IMO, it's the lenses that need the most attention. Forget about resolution for a while - 100 and incoming 150 is already mostly unnecessary for the time being.
It's pretty clear that the abilities of the sensors are outstripping those of the computers in the cameras. And of course the sensors improve every year, the lenses slowly.
One example is DR: Put a camera indoors in front of a window, and image a vase of flowers against the light. The sensor may be able to do it, but the lens will flare out your photo's indoor details.
Edmund
Edmund
-
Where Medium Format main players have failed is that they've piled all their money into backs and sensors and in the case of Phase One, the XF Body. But the blue rings lenses aren't massively better, they are the same optical designs as before just tarted up. Hasselblad lenses are decent but again, they are aging and aren't looking very good next to the best 35mm lenses.
IMO, it's the lenses that need the most attention. Forget about resolution for a while - 100 and incoming 150 is already mostly unnecessary for the time being.
But then H lenses were working on the 200 mpixels multishot already a few years ago.
-
But then H lenses were working on the 200 mpixels multishot already a few years ago.
There is a difference between working and wowing. Stopped down the lenses are good. No one shoots with the Multishot camera wide open.
Like I said, the H lenses are very good but next to the best 35mm lenses they aren't so strong.
-
Hi,
Most H-lenses are pretty good, as far as I can recall. But, a lot of things have been going on in lens design and it seems that the lenses for the GFX are playing in another division.
Generally, when you scale down a lens design it will get higher resolution. Cell phone lenses have extremely high resolution, in lp/mm but the sensors are also small.
The major issue with cell phones is probably that they are thin. That limits the length of the lens.
Much of the development of sensors originates in small sensors and migrates to larger sensors. I don't think cell phone cameras ever will catch MFD, but I would think they will be good enough for most things.
On the other hand, I would think that once you mount the camera on a tripod it may as well be a more substantial one.
Best regards
Erik
-
There is a difference between working and wowing. Stopped down the lenses are good. No one shoots with the Multishot camera wide open.
Like I said, the H lenses are very good but next to the best 35mm lenses they aren't so strong.
Your message implied that nobody needed the increased resolution. I noticed that some photographers used a multishot camera because they did. Your message also implied that lenses were not good enough. I noticed they were good enough for multishot users. True: they may have needed to stop the lens down (a bit), but that is not necessarily a problem.
FWIW, at next Photokina you will get cameras with the 150 mpix sensor and no new lenses in H mount, so the discussion is settled.
-
Most H-lenses are pretty good, as far as I can recall. But, a lot of things have been going on in lens design and it seems that the lenses for the GFX are playing in another division.
If I remember correctly, you do not own H lenses (you own CF lenses) and neither do you own a GFX and its lenses. I would bet that you never rented or borrowed them to try them either.
-
Your message implied that nobody needed the increased resolution. I noticed that some photographers used a multishot camera because they did. Your message also implied that lenses were not good enough. I noticed they were good enough for multishot users. True: they may have needed to stop the lens down (a bit), but that is not necessarily a problem.
FWIW, at next Photokina you will get cameras with the 150 mpix sensor and no new lenses in H mount, so the discussion is settled.
No, the discussion isn't settled.
I did imply that in some ways they aren't good enough because they aren't. That's coming from some one who has used them for a decade.
Just because they work, and will continue to work OK on 150MP, doesn't mean they work perfectly in every way and doesn't mean they don't have flaws that you have to work around. It certainly doesn't mean they don't need improving to bring them up to scratch with long overdue updates.
The newer lenses like the 50 II and the 120 II are much better.
-
Your message implied that nobody needed the increased resolution. I noticed that some photographers used a multishot camera because they did. Your message also implied that lenses were not good enough. I noticed they were good enough for multishot users. True: they may have needed to stop the lens down (a bit), but that is not necessarily a problem.
FWIW, at next Photokina you will get cameras with the 150 mpix sensor and no new lenses in H mount, so the discussion is settled.
A few things to keep in mind.
First, the advantage of a multi-shot camera to a professional is not the resolution, but the fact that you are not relying on interpolation to determine the two missing RBG values for each pixel. (Remember, each pixel only sees either red, blue or green, but each pixel needs a value for all three to create a color image). This cuts down greatly on color artifacts, such as color morie and digital blooming.
Second, cutting out interpolation created sharper edges on objects too. This is complicated to explain why, but the results are striking, and certainly made lenses appear better then they were.
Third, of course the added resolution does add a benefit as well, but not in the way you are thinking about it. In the early days of digital, pattern morie was more of an issue too, and increasing resolution decreased this as well. This is extremely useful when shooting speakers and objects with very small repetitive patterns.
Last, even though the multi-shot backs gave you more resolution, the pixels were still relatively large. Each of the 4, or 6, shots per digital capture were using the same size pixels as a single shot capture, so the lenses did not need to resolve any better then for the a single shot capture of 39 MP.
-
Hi,
On the other hand, I know how to read MTF data and both Hasselblad and Zeiss publishes MTF data.
I also know a guy, Jim Kasson, who owned quiet a few V and H-series lenses and did very detailed comparison between those lenses and native lenses for the GFX.
Jim sold off all his Hasselblad H and V gear.
Hasselblad did some detailed comparisons between some of the HC lenses and the CF/CFi/CFE lenses when the H cameras series were released. In that comparison that Hasselblad made, the HC-lenses had a more balanced comparison between infinity and close range, but near infinity the lenses were pretty close: http://static.hasselblad.com/2015/02/the_evolution_of_lenses.pdf
You may think that I can lease a lens and camera, with unknown history and use it for some days and get more reliable results than data measured by Hasselblad using gear costing up to something like a million $US operated by the designer of the HC-lenses?
Just to say, Hasselblad publishes MTF data for both HC-series and X-series, why don't you download and check?!
Best regards
Erik
If I remember correctly, you do not own H lenses (you own CF lenses) and neither do you own a GFX and its lenses. I would bet that you never rented or borrowed them to try them either.
-
Just because they work, and will continue to work OK on 150MP, doesn't mean they work perfectly in every way and doesn't mean they don't have flaws that you have to work around. It certainly doesn't mean they don't need improving to bring them up to scratch with long overdue updates.
I never said they were perfect. No physical system is.
The newer lenses like the 50 II and the 120 II are much better.
Indeed they are.
-
Last, even though the multi-shot backs gave you more resolution, the pixels were still relatively large. Each of the 4, or 6, shots per digital capture were using the same size pixels as a single shot capture, so the lenses did not need to resolve any better then for the a single shot capture of 39 MP.
I don't think this is mathematically correct. You are sampling at double the resolution with a larger window.
-
I don't think this is mathematically correct. You are sampling at double the resolution with a larger window.
The increase in resolution is from using 4, or 6, shots at 39 MP each to upscale to a single larger file size. Although it is not 100% similar to upscaling from a single capture, it is still relies on multiple native resolution of 39 MP shots to get the job done. Therefore the resolving power of the lens does not need to be any better for a higher resolution multi-shot image then it needs to be for the native 39 MP single shot image.
On top of this, there is no interpolation being used here. I did not want to explain this before, but will now. Since each pixel only sees either red, green or blue in a single capture, the computer uses the neighboring pixels of opposite color to the one in question to interpolate, or guess, the missing two values for that pixel, for every single pixel. This presents a number of problems, one in particular is for pixels at a sharp edge in images where there are two drastically different shades on either side.
Here, since each pixel at or near the edge is having its missed color values based upon pixels on both sides of this border, the computer does not exactly know how to proceed. So, depending on how drastic the shades are, the interpolation will create graduated values for these missing colors and will, in effect, create a slighting blurred edge.
With a multi-shot capture, the sensor moves one pixel per shot, so the computer knows the exact value for all three colors for each pixel. There is no guess work involved, thus the problem described above does not occur (and a few others too).
Now, typically, most photographers will not even notice this, even at 100%, since most have not seen a single capture of the same image compared to a multi-shot capture. But, if you ever do, it is very noticeable and makes the image look sharper at 100%.
As a matter of fact, currency is designed around the blurring created from interpolation to deter counterfeiters. Fine details in money often do not show up in the image, or not sharp enough to print from. With a multi-shot capture, all the detail is present.
So, because of this, even an okay lens on a multi-shot capture will make a sharper looking image, due to no interpolation at work, then a great lens on a single shot capture.
Therefore, we can not make assessments about the quality of lenses based on multi-shot images for single shot captures that are approaching the same resolution of that multi-shot capture.
-
On the other hand, I know how to read MTF data and both Hasselblad and Zeiss publishes MTF data.
We were talking about GFX lenses. Fuji publishes MTF for those, but the curves lack units and do not specify the aperture. Even then, a comparison of the GFX 23mm and the HCD 24mm appears to show that the H lens is at least as good.The GFX 45mm would be a bit better than the HC 50-II, but only in the center. OTOH, the GFX 110 and Macro 120mm lens would be vastly better than anything on this planet, if the curves are to be believed.
I also know a guy, Jim Kasson, who owned quiet a few V and H-series lenses and did very detailed comparison between those lenses and native lenses for the GFX.
I have read his blog.
-
The increase in resolution is from using 4, or 6, shots at 39 MP each to upscale to a single larger file size. Although it is not 100% similar to upscaling from a single capture, it is still relies on multiple native resolution of 39 MP shots to get the job done. Therefore the resolving power of the lens does not need to be any better for a higher resolution multi-shot image then it needs to be for the native 39 MP single shot image.
Nope. The larger pixel aperture indeed adds a low-pass filter, but does not suppress high-frequency content. The reconstruction process can extract the higher frequency content from the data.
If your theory was correct, there would be no difference between the multishot mode giving the same number of pixels (which indeed simply suppresses colour aliasing) and the mode where additional samples are taken by moving in half-pixel steps. Yet, the two modes exist and give different results. Actually, Hasselblad sells two cameras: a 50MS (just removes colour aliasing) and 200MS (removes colour aliasing and increase resolution by taking extra pictures using half steps).
-
Nope. The larger pixel aperture indeed adds a low-pass filter, but does not suppress high-frequency content. The reconstruction process can extract the higher frequency content from the data.
If your theory was correct, there would be no difference between the multishot mode giving the same number of pixels (which indeed simply suppresses colour aliasing) and the mode where additional samples are taken by moving in half-pixel steps. Yet, the two modes exist and give different results. Actually, Hasselblad sells two cameras: a 50MS (just removes colour aliasing) and 200MS (removes colour aliasing and increase resolution by taking extra pictures using half steps).
Sure, this is correct, but the captures are still 50 MP each. So, the lens does not need to resolve for a 200 MP capture; it just needs to be good enough for a 50 MP capture.
The half steps are simply used to give the computer a much better guess on how to create additional details during the upscaling process. We have to assume there is certain amount of information between each pixel that is not being captured, just due to the space that exists between each light absorbing diode. The half steps solves this problems, to a degree, and allows the computer to see the missing information.
Kind of like reading between the lines, literally.
So, I say again. If a lens works well on a 200 MP image produced from a 6-shot 50 MP multi-shot capture, this does not mean that same lens will work well with a 200 MP single shot capture.
-
Sure, this is correct, but the captures are still 50 MP each. So, the lens does not need to resolve for a 200 MP capture; it just needs to be good enough for a 50 MP capture.
I don't agree. The details need to be transmitted by the lens. The many samples taken after the lens can be of low resolution, enough of them will allow to reconstruct the data transmitted by the lens but not more.
Actually,a similar oversampling process is used in astronomy: you take lots of pictures after a telescope to increase resolution. You are still limited by the resolution of the telescope. An article: http://www.adass.org/adass/proceedings/adass99/O6-02/
-
I don't agree. The details need to be transmitted by the lens. The many samples taken after the lens can be of low resolution, enough of them will allow to reconstruct the data transmitted by the lens but not more.
Actually,a similar oversampling process is used in astronomy: you take lots of pictures after a telescope to increase resolution. You are still limited by the resolution of the telescope. An article: http://www.adass.org/adass/proceedings/adass99/O6-02/
I think this is an apples to oranges comparison.
With astronomy it is not practical to use multi-shot technology where the sensor is moving a pixel each time, allowing for sampling a of each color value for each. The earth is moving, along with whatever is being observed. So here, even though multiple captures are being use, the computer does not know the color value for each pixel and is still guessing. This is partly different then what Hassy does by eliminating the interpolation.
Also, multiple captures of the night sky are not only used to increase resolution, but to also decrease noise. Dark images shot at a higher then native ISO is going to create a far amount of noise with a ccd sensor. Having many captures of (about) the same scene helps at removing that noise.
So maybe the lens does have an effect here. However, I still think, due to a lack of interpolation, we can not assume a lens that worked well with a 200 MP multi-shot capture will work with a 200 MP single shot one that utilizes interpolation.
-
I had a chance to look at some comparative test shots from the 400MS today - they weren't that well taken, but there was enough there to gleam an idea of what it could do: the greater resolve in detail gained by the RGBG Multishot in comparison to the Single shot was plainly evident: just the same as when comparing the 22MP/39MP/50MP Multishot files with their Single Shot counterparts. However when comparing the 6S files with Multishot file, less extra detail was resolved in comparison with what the 200MS 6S was able to do in comparison with its Multishot counterpart. Simply speaking I felt that I was looking at a high resolution image of the interior of the HC120II and the pixel grid was way finer than the tiniest bit of image detail.
Sent from my iPhone using Tapatalk
-
New PromoVideo
https://youtu.be/xr-FjdyXZhQ
Info site
http://www.noa-mobile.hr/element/n7
-
Well for me its lighting, moment or atmosphere, or a combination of all three that decides if a picture is any good. Pixels don't count a heap to anyone other than photographers.
I wonder what resolution Canaletto painted at, anyone ever look at his paintings and say "if only we could see more bricks"? lets face it they could paint to any level of detail they thought necessary, but enough was and is enough :-)
-
Well for me its lighting, moment or atmosphere, or a combination of all three that decides if a picture is any good. Pixels don't count a heap to anyone other than photographers.
Male photographers Kevin. Male photographers....
(It's a bit the same buzz as the dudes who tune their wrecks with chrome and dragster exhausts to show muscles and impress the next door farm's redneck striking blonde. Instead of horsepowers and noise, we have megapixels and lack of noise.)
-
Well for me its lighting, moment or atmosphere, or a combination of all three that decides if a picture is any good. Pixels don't count a heap to anyone other than photographers.
I wonder what resolution Canaletto painted at, anyone ever look at his paintings and say "if only we could see more bricks"? lets face it they could paint to any level of detail they thought necessary, but enough was and is enough :-)
I don't know about Canaletto, but I went to a gallery in Paris that was selling a Breughel collection, a roomful of them, and the Breughel the Younger pieces definitely rewarded viewing with the loupe which they courteously provided.
Edmund
-
I don't know about Canaletto, but I went to a gallery in Paris that was selling a Breughel collection, a roomful of them, and the Breughel the Younger pieces definitely rewarded viewing with the loupe which they courteously provided.
Edmund
Like I said they could paint to any level they thought it needed, Breugel painted detail where he thought he needed to. But that isnt what makes a Breugel compelling. Chances are he has used light and atmosphere to capture a past Flemish scene. The point is the difference between 30mp and 100mp isnt going to be what makes an image good or bad, the constant search for more blades of grass in an image is only something photographers bother about, society will not pick an image on detail no more than they give a dogs dangler about Bokeh. There is a much greater list before you get down to splitting hairs on less important details. More pixels sharper lenses is an easy fix for a photographer that thinks they can make what they have better and its an easy idea to sell.
I think the largest selling picture recently has been a couple doing the Tango on the beach with the butler holding an umbrella, very little detail but lots of atmosphere.
Cheers,
Kev
-
Hi,
Most H-lenses are pretty good, as far as I can recall. But, a lot of things have been going on in lens design and it seems that the lenses for the GFX are playing in another division.
Best regards
Erik
Hi,
Yes same as the X1D lenses, as illustrated here: https://www.youtube.com/watch?v=wD3An09hmjU
Hasselblad was updating the H series lenses, now, with the focus on the X1D, I am not sure if the process continues.
One don't nesesary want hyper corrected lifeless lenses, but we need a little more resolution.
I hope it does.
Best regards,
-
It looks like those millions of dollars of R&D invested by phone makers are starting to pay off ...
https://www.dpreview.com/news/2999078463/noa-n7-smartphone-captures-80mp-images-with-high-resolution-mode
Edmund
Yes, Not sure about 80MP, but Huawei is releasing a phone with Triple rear camera - 40+20+8MP camera sesnor on rear side and 24MP front sidee.
That is upcoming Huawei P20 Pro - Specs are not comfirmed yet.
You can check full details here - http://techenroll.com/aps-product/huawei-p20-pro/ (http://techenroll.com/aps-product/huawei-p20-pro/)