3 ms·

The HDRI shot on top of the site is very nice! But I don't understand why these shots are still not possible with current camera's. If you take a picture with t

by tintin 15y ago

The HDRI shot on top of the site is very nice! But I don't understand why these shots are still not possible with current camera's. If you take a picture with the longest exposure time wouldn't it be possible to dim the brightest areas using a formula?

sp332 15y ago
If you overexpose that part of the sensor, it simply records the maximum value. Then when you "dim" it later, you just get a blank grey area. In order to extract detail, you have to maintain some contrast between "bright" and "really bright", so you have to keep the "bright" area below the maximum value of the sensor.
- tintin 15y ago
  But doesn't a camera sample pixels at a given rate? So lets say a camera can access image data from the sensor @ 100Hz and you store this over time (exposure time) it doesn't matter how bright areas are. But I can be completely wrong. Maybe I'm confusing frame-rate (video ability) with the sensor burst-rate.
  - artmageddon 15y ago
    I'm not entirely sure what you mean by sampling pixels, but the parent's post is saying that some areas are very bright while some areas are very dark, like you see in the HDR shot.. but the sensor doesn't know which of the two light levels you want to adjust it for, so it makes a best-guess based on your camera settings. Example: while our eyes are much more advanced than any camera; think of the effect of having a giant spotlight shined directly* into your eye: You can see the light, but everything else is going to be much darker in comparison. If you look at it long enough, eventually all you're going to see is just the spot-light, everything else around you will be dark. Much like your eye, this is exactly what would happen to your picture if you make the exposure time long enough.. and it doesn't take very long, even 0.25 or 0.5 seconds of time can ruin a picture taken outdoors. To answer your original post: a lot of DSLRs nowadays DO have features that do this, called "Bracketing" mode where it'll take the same picture with low, mid-range, and high exposure levels(hence the "high dynamic range" name). From there, you combine the three pictures(or more) in photoshop or other software and you get that resulting image that you see. In the low mode, you'll be able to see the foreground while the background is mostly white, the middle level will have both areas show but with not-too-pleasing colors, while the high level will show the background, but the foreground will appear completely dark. *personal experimentation from long ago has shown that this isn't advisable :) Edit: I should add that bracketing is just a special feature that takes the pain out of having to set the exposure time and aperture(how wide open the lens is when shooting), but these kinds of HDR shots are perfectly capable with any camera(yes, even the pocket-sized ones) if you have control over these settings. Knowing how to use the camera is key(along with keeping it perfectly still).
    - tintin 15y ago
      "I'm not entirely sure what you mean by sampling pixels" I think I'm confusing some things. I can understand that each pixel of a sensor can output values from lets say 0 to 1V where 0V is dark, 1V is light. So you need a shutter-time because else the sensor will only output values of 1V. I was thinking todays cameras can sample these values at a very high frame-rate. So when you sample these values over time and store them you can expand the range. But it seems like "bracketing" is doing something like this. I did not know. Thanks for the info!
      - grecy 15y ago
        I've always found this interesting too, especially when a RAW file grows in size with exposure time... doesn't that indicate it is "sampling" the sensor?

by tintin 15y ago

sp332 15y ago

If you overexpose that part of the sensor, it simply records the maximum value. Then when you "dim" it later, you just get a blank grey area. In order to extract detail, you have to maintain some contrast between "bright" and "really bright", so you have to keep the "bright" area below the maximum value of the sensor.

tintin 15y ago
But doesn't a camera sample pixels at a given rate? So lets say a camera can access image data from the sensor @ 100Hz and you store this over time (exposure time) it doesn't matter how bright areas are. But I can be completely wrong. Maybe I'm confusing frame-rate (video ability) with the sensor burst-rate.
- artmageddon 15y ago
  I'm not entirely sure what you mean by sampling pixels, but the parent's post is saying that some areas are very bright while some areas are very dark, like you see in the HDR shot.. but the sensor doesn't know which of the two light levels you want to adjust it for, so it makes a best-guess based on your camera settings. Example: while our eyes are much more advanced than any camera; think of the effect of having a giant spotlight shined directly* into your eye: You can see the light, but everything else is going to be much darker in comparison. If you look at it long enough, eventually all you're going to see is just the spot-light, everything else around you will be dark. Much like your eye, this is exactly what would happen to your picture if you make the exposure time long enough.. and it doesn't take very long, even 0.25 or 0.5 seconds of time can ruin a picture taken outdoors. To answer your original post: a lot of DSLRs nowadays DO have features that do this, called "Bracketing" mode where it'll take the same picture with low, mid-range, and high exposure levels(hence the "high dynamic range" name). From there, you combine the three pictures(or more) in photoshop or other software and you get that resulting image that you see. In the low mode, you'll be able to see the foreground while the background is mostly white, the middle level will have both areas show but with not-too-pleasing colors, while the high level will show the background, but the foreground will appear completely dark. *personal experimentation from long ago has shown that this isn't advisable :) Edit: I should add that bracketing is just a special feature that takes the pain out of having to set the exposure time and aperture(how wide open the lens is when shooting), but these kinds of HDR shots are perfectly capable with any camera(yes, even the pocket-sized ones) if you have control over these settings. Knowing how to use the camera is key(along with keeping it perfectly still).
  - tintin 15y ago
    "I'm not entirely sure what you mean by sampling pixels" I think I'm confusing some things. I can understand that each pixel of a sensor can output values from lets say 0 to 1V where 0V is dark, 1V is light. So you need a shutter-time because else the sensor will only output values of 1V. I was thinking todays cameras can sample these values at a very high frame-rate. So when you sample these values over time and store them you can expand the range. But it seems like "bracketing" is doing something like this. I did not know. Thanks for the info!
    - grecy 15y ago
      I've always found this interesting too, especially when a RAW file grows in size with exposure time... doesn't that indicate it is "sampling" the sensor?