Measuring 7D noise vs. ISO and HTP

Started Aug 28, 2012 | Discussions thread
Flat view
Thiago Maia New Member • Posts: 12
Measuring 7D noise vs. ISO and HTP

So I decided to check for myself the effects of ISO and HTP on noise. I've been shooting on 160 ISO multiples with the 7D for the last couple of years based on several posts out there. I guess this is yet another post to confirm these findings.

This is the little experiment I made:

  • shot my 7D (FW 2.0.0) without lens (lens mount cap on) under a blanket in a dark room (just to be sure there would be hardly no light hitting the sensor) at 1/8000s

  • shot with different combinations of ISO and HTP settings, the beginning of the ISO scale in 1/3 stops, full stops after ISO 400

  • shot 3 times with each setting (ok, could have done more samples per setup for better statistical significance, but the standard deviations were low and it would have taken far too long to process)

  • all shots taken in RAW were batch-processed in DPP 3.11.31 with RAW brightness adjustment set to +2.0, all other settings set to defaults

  • all shots exported to 16-bit TIFFs

  • each TIFF processed with a custom script in Matlab (scientific programming software) to calculate a simplified measure of entropy

The entropy measure of an image may be interpreted as the amount of information carried by that image. As there was (almost) no external stimuli to the sensor, it was fair to assume that whatever the sensor was capturing was a result of the internal noise generated by signal amplification, anti-alias filter, circuitry, algorithms, etc. Therefore, the entropy of the captured images with different ISO and HTP settings would give us a quantitative measure of noise generated by camera itself.

Results are in the image attached. First, 160 and 320 did display less noise than 100 and 200 respectively. This confirms so many other reports found in different forums. I'll try to shoot the whole scale in 160 multiples later for a broader view of the effect as this time I only focused in the beginning of the scale. Second, HTP does increase noise significantly. We read of this effect often, but I didn't expect it to be so much. HTP-ON+200 has more noise than HTP-OFF+400!

Now that the experiment is over (and took a couple of hours to process batch+scripts 60 110Mb TIFF images!), I regret not having shot the entire scale for a better comparison of the 160 ISO multiples. Nevertheless, it does appear that in practical terms it is better to go for 160 multiples rather than 100 multiples, at least in the beginning of the scale.

As for HTP, I often just left it ON depending on the type of shoot. Now I will use it with a little more care. More DR in bright areas is of course nice, but it's good to understand the price you pay in the shadows. In addition, I used to shoot HTP-ON+320 based on the assumption that 160 multiples would be better than 100 multiples with HTP-ON. It seems this effect is not as pronounced as with HTP-OFF, leading HTP-ON+320 to have more noise than HTP-ON+200. Also, for high ISO (above 1600), HTP-ON apparently is as noisy as HTP-OFF.

Quick DIY: here is a quick test that takes less than a minute with rather surprising results (for me at least!). Shoot RAW with the lens cap on, in a dark place, with ISO 6400 or 12800, 1/8000, f/22 (or smaller). Then check the image in DPP and bump RAW brightness by +2.0. View the image at 100%. I didn't expect to see so much noise over a black image background.

C&C more than welcome.

Best regards, Thiago

Canon EOS 7D
If you believe there are incorrect tags, please send us this post using our feedback form.
Flat view
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow