Large Sensor vs Computational Photography
a sort of Mind versus Matter battle but for cameras !
smartphone night mode
“photorealism is dead, photos taken on modern smartphones are artificial intelligence hallucinations”
Why is he sitting at a desk looking like a student in detention?
Anyway, the idea the video is presenting is interesting.
The simulated picture is to be preferred over what is there.
We prefer the illusion to reality.
And the “faker” it is, the better it is.
What’s next? Eye implants?
1 Like
WIRED is an establishment outfit. the presenter is brown and his guest is also brown. the girl he takes a picture of is also non-white. he’s basically telling his audience that the future is non-white and SLAVE. sit down and shut up. speak when spoken to. know your place. and of course he gets lectured ( while sitting at a tiny school desk ) by a PROFESSOR and the girl is wearing a MASK while he is taking a picture of her !
i see your point and in fact the professor in the video made very similar points and i agree with both of you, namely that people don’t want to remember things as they really were but as they wish they were …
i could cite examples using my parents but i don’t want to go there so instead i will use an example from the ( otherwise quite bad ) movie “Buffalo 66” in which “Billy” comes out of prison and goes to see his parents and his mother offers him chocolate cake and he reminds his mother that he is so allergic to chocolate that he almost died the last time he accidentally had some and his mother replies " what you are you talking about you love chocolate ! " His mother’s “memories” are nothing but her own fantasies, completely disconnected from reality. My own mother is exactly the same way, of course.
but this isn’t where i was trying to go with this thread …
instead the point is early digital cameras were really just film cameras that replaced film with an image sensor … whereas future digital cameras will be more like human vision …
human vision happens mostly in the brain, not the eye … and that’s what computational photography is about …
our eyes actually are very limited as sensors - we only have sharp vision at the very center of our field of vision and in fact there are completely blind spots as well but we aren’t aware of any of this because our brain masks this from us by moving the eyes and using memory and imagination …
what we think we “see” is actually our imagination. we actually “see” very little. when the system works well the experience is completely seamless. when people have dementia or are using hallucinogenic drugs they begin seeing things which aren’t there. we also tend to see faces in everything and there are other optical illusions as well.
all these things are simply evidence that what we “see” isn’t actually real.
computational photography is really about making cameras work the way our vision does - by leaning on processor and software to recreate the scene from very limited amount of information that a tiny smartphone imaging sensor can read out at any one time …
for example the reason you can so easily read this text is because your mind already contains the images of all the letters and even entire words, as well as expressions and sentence structures … from time to time you might read something other than what i wrote because you aren’t actually reading my words - your brain is writing its own text inspired by what you’re seeing on the screen.
computational photography isn’t that advanced yet but it’s the future. whereas 20 years ago it was all about improving the sensors and sensor quality may still be important for video today … for still images it is now about what you do with the information from the sensor, rather than the sensor itself.
so in the past still cameras had much larger sensors than video cameras because video cameras captured information from the sensor continuously whereas still cameras captured information in a single millisecond and thus the sensor had to be very powerful to capture meaningful information …
but in the future still and video camera sensors will be the same and still cameras will capture information continuously just as video cameras do and then computational techniques will generate still images from that continuous stream …
smartphones are already starting to do this out of necessity ( they can’t use large sensors ) while big professional cameras are still using dumb technology and gigantic sensors because they can …
but the customer is abandoning large cameras in favor of smartphones and this is pushing professional camera makers to start experimenting with computational photography as well instead of simply relying on gigantic lenses as in the past …
Olympus OM-1 is an example of a camera that is trying to bring smartphone computational photography technology to professional camera market but in a smaller and lighter form factor than the industry leading Sony and Canon mirrorless cameras …
so far the results from Olympus OM-1 are mixed and in most cases the larger Sony and Canon will still deliver better pictures, but the real pain point is that even though OM-1 is able to take 120 pictures per second using traditional method it takes SEVEN SECONDS to take a single image using computational photography …
in other words even though OM-1 has one of the fastest processors in a camera it’s still very slow compared to a phone processor let alone a human brain …
never the less it is quite clear where the future lies - it isn’t in better sensors - but in faster processors - and FASTER ( as opposed to sharper ) sensors as well …
the speed of OM-1 that allows it to take 120 pictures per second doesn’t just come from fast processor but also from fast SENSOR …
OM-1 uses a STACKED sensor which is a new high-end technology that allows for much higher sensor readout speed … Sony originally developed this technology to enable things like 8K video, which require extremely fast sensor readout but in OM-1 the technology is used to INTERNALLY take multiple shots in rapid succession which are then combined into a single shot using computational photography …
so in fact the 120 shots per second ability that OM-1 advertises for sport photography ( if you want to capture the exact moment the bat hits the baseball for example you would take a 120 frame per second burst and then later choose the shot which captures the right moment ) is really a welcome side-effect but the real reason the camera needed this speed was for computational photography features …
so you can see how still photography and video are starting to merge where still cameras of the future will have smaller but faster sensors just like video cameras were already using … this is really interesting for me because of course i want a camera that is a hybrid, able to take both good stills and good video … and computational photography is what is going to make it possible …
OM-1 will take 8 to 12 shots in a fraction of a second and then use another ~ 7 seconds to process them into a single shot … thus the SPEED of both the sensor and processor become critical while the native image quality from the sensor is starting to become less important …
DP Review only gave OM-1 a SILVER award ( not gold ) because it is quite expensive and the image quality is not as good as Sony or Canon … but Sony and Canon require much larger, heavier and more expensive lenses to achieve their image quality …
if you’re a professional the consensus in 2022 is to go with Sony with many people switching from Canon to Sony in the last few years due to Sony having better support for Video than Canon does …
but if you’re looking for a semi-professional system more suitable for travel due to being cheaper, lighter, more compact and better weather sealed then Olympus OM-1 is a strong contender in 2022 …
i actually have an Olympus DSLR but it’s very old and doesn’t take videos at all. theoretically if i could find it i should be able to use my old lenses on the new OM-1 body with an adapter as they both use the same image sensor size but i never really used my old DSLR that much even when phone cameras were worthless so would i really use a new one now that my phone already takes good pics ?
for a while i wanted to get a good video-capable semi-pro camera for my YouTube channel but i have been mulling deleting my YouTube channel lately because i don’t want to keep doxxing myself.
sorry this was a very long post.