Pls show examples where increasing brightness in post better than increasing ISO.

Started Oct 1, 2013 | Discussions thread
Jeff Veteran Member • Posts: 5,568
Re: The meaning of ISO, NEX-7 vs. EOS 7D

ProfHankD wrote:

Jeff wrote:

PhilPreston3072 wrote:

With the latest sensors, some have been proclaiming it's better to shoot at base ISO and increase brightness in post rather than increase ISO in camera.

I would like to see some examples of this. Are there any examples where increasing brightness in post delivers better IQ than increasing ISO in camera? And please, disable all NR settings.

Here's a comparative example from Pierre Sottas comparing the same exposure made at ISO 100 and ISO 6400. Note the difference in how the highlights have been captured.

ISO film speed is a bit of a judgement call for sensors used in digital cameras. Basically, the notion of ISO change is implemented by two things:

  • Change of the exposure settings (aperture and shutter speed)
  • Change of the sensor gain

The comparison being made in is keeping the exposure settings of the higher ISO, just comparing changing the sensor gain in camera (which is presumed to be a change in analog amplification) with changing the gain in digital post-processing. Depending on noise/dynamic range in the values out of the ADC, the raw difference can be tiny. The Magic Lantern folks have repeatedly claimed that Sony-made sensors are "ISO-less" in that there is very little difference between the two methods for changing gain, but their dual ISO hack for several Canon DSLRs shows a claimed 3-stop improvement in dynamic range by setting the analog gains differently for the two interleaved sensor readout channels.

I think what's going on is that Sony's sensors have exceptionally low noise in their analog processing -- for example, compare the DxOMark scores of the NEX-7 and EOS 7D. Color depth is 24.1 vs. 22 bits and dynamic range is 13.4 vs. 11.7. In short, the Sony sensor is about 2 stops "cleaner" as the data comes off the sensor at low ISO. Even more interesting is the dynamic range curve comparison of the two cameras as ISO setting is changed. Dynamic range decreases for both cameras as ISO is increased, generally decreasing about 1 stop per stop increase in ISO -- which would mean increasing ISO basically adds nothing. However, the Canon sensor doesn't decrease dynamic range much until ISO 400 -- whereas the Sony sensor drops steadily, but isn't as bad as the Canon at 100 until it reaches 400. Both sensors perform identically at 800.

What does all that mean? Basically, it says that the claim of Sony sensors being ISO-less is well justified (although there is a little bump in the dynamic range that gives about 1/2 stop improvement at 6400, probably due to raw noise reduction kicking in). The Canon sensor, however, just doesn't have enough analog headroom to get all the dynamic range the lowest few ISOs could have provided, so it can get somewhat more than 2 stops improvement by bumping the ISO to something around 400 or 800.

Anyway, this is an interesting phenomenon, and perhaps I'll have the students in my "Cameras as Computing Systems" course do a little project on this....

There are others that can speak much more authoritatively about in-camera analog versus digital processing of the sensor data. And I'm not a Canon shooter so can't say much about that, either. But by simple E-P5 behaves as one would expect from iso-less perspective, which so far has been good enough for me.

Sound's like an interesting course.  With that source of 'free labor', it might be fun to do some controlled tests under typical shooting conditions.

I assume you've seen the sensor gen site

 Jeff's gear list:Jeff's gear list
Olympus PEN E-P5 Olympus E-M1 Olympus OM-D E-M1X Olympus Zuiko Digital ED 150mm 1:2.0 Olympus M.Zuiko Digital 45mm F1.8 +6 more
Post (hide subjects) Posted by
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow