Home · Register · Join Upload & Sell

  

  Previous versions of cgarcia's message #13468978 « Canon 80D dynamic range: good news! (quick test with RAW images) »

  

cgarcia
Offline
Upload & Sell: Off
Canon 80D dynamic range: good news! (quick test with RAW images)


I have completed a more detailed study of the RAW images. Definitely, the sensor dynamic range versus ISO graph now resembles the profile of the Canon competitors. To make this graph I have used the 7D2 results generated with exactly the same technique on september 2014 (see https://www.fredmiranda.com/forum/topic/1319060/0):







However, at high ISO, the 80D is slightly under the 7D2 level. I don\'t know if this is related to cheaper electronics or if the new tech has some influence. Note also that both cameras true ISO level could differ.

Here are the numbers, using also the ones from my 7D2 analysis (same technique) to compare:

ISO 100 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00100NR0.CR2.HTM)
80D - DR 13.22 at 8MP (12.43 at 24MP), read noise 2.8712 DN
7D2 - DR 11.60 at 8MP (10.94 at 24MP), read noise 5.88731 DN

ISO 200 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00200NR0.CR2.HTM)
80D - DR 12.74 at 8MP (11.95 at 24MP), read noise 3.99545 DN
7D2 - DR 11.70 at 8MP (11.04 at 20MP), read noise 6.28073 DN

ISO 400 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00400NR0.CR2.HTM)
80D - DR 11.92 at 8MP (11.13 at 24MP), read noise 6.37543 DN

ISO 800 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00800NR0.CR2.HTM)
80D - DR 11.09 at 8MP (10.29 at 24MP), read noise 11.3868 DN

ISO 1600 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI01600NR0.CR2.HTM)
80D - DR 10.25 at 8MP (9.45 at 24MP), read noise 20.3616 DN
7D2 - DR 10.41 at 8MP (9.75 at 20MP), read noise 15.4014 DN

ISO 3200 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI03200NR0.CR2.HTM)
80D - DR 9.41 at 8MP (8.62 at 24MP), read noise 36.395 DN

ISO 6400 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI06400NR0.CR2.HTM)
80D - DR 8.63 at 8MP (7.84 at 24MP), read noise 62.2559 DN
7D2 - DR 8.95 at 8MP (8.29 at 20MP), read noise 42.3891 DN

ISO 12800 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI12800NR0.CR2.HTM)
80D - DR 7.64 at 8MP (6.85 at 24MP), read noise 124.027 DN
7D2 - DR 8.00 at 8MP (7.34 at 20MP), read noise 82.8706 DN

ISO 16000 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI16000NR0.CR2.HTM)
80D - DR 7.33 at 8MP (6.54 at 24MP), read noise 153.517 DN

ISO 25600 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI25600NR0.CR2.HTM)
80D - DR 6.74 at 8MP (5.95 at 24MP), read noise 231.801 DN

The new sensor uses a black level of DN (data number) 512 for ISO 100 and 200, but continues using 2048 for the remaining ISO levels. The saturation level seems to be the DN 16383 at all ISO levels, which is unusual. For example, the analyzed ISO 100 picture has 871 blue pixels with DN 16383, 1039 red pixels with DN 16382, and 1660 plus 1632 green pixels (two green channels) with DN 16383; the image is overall underexposed and from DN 10000 to 16381/2, each DN only counts 0-2 pixels.

There are two masked areas in the 25.5MP image (6288x4056): one 264 pixels wide at the left and the other 34 pixels height at the top. So, the effective image area is 6024x4022 (one-two pixels in the frontier of both areas, showing visible artifacts, are not truly usable). The masked area does not receives light and its standard deviation can be measured to determine the read noise, and to infer from that the DR. The area used has been the left one, but the top measures almost the same noise levels (unlike with other sensors, where the top masked area has some oddities).



Mar 17, 2016 at 08:44 PM
cgarcia
Offline
Upload & Sell: Off
Canon 80D dynamic range: good news! (quick test with RAW images)


I have completed a more detailed study of the RAW images. Definitely, the sensor dynamic range versus ISO graph now resembles the profile of the Canon competitors. To make this graph I have used the 7D2 results generated with exactly the same technique on september 2014 (see https://www.fredmiranda.com/forum/topic/1319060/0):







However, at high ISO, the 80D is slightly under the 7D2 level. I don\'t know if this is related to cheaper electronics or if the new tech has some influence. Note also that both cameras true ISO level could differ.

Here are the numbers, using also the ones from my 7D2 analysis (same technique) to compare:

ISO 100 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00100NR0.CR2.HTM)
80D - DR 13.22 at 8MP (12.43 at 24MP), read noise 2.8712 DN
7D2 - DR 11.60 at 8MP (10.94 at 24MP), read noise 5.88731 DN

ISO 200 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00200NR0.CR2.HTM)
80D - DR 12.74 at 8MP (11.95 at 24MP), read noise 3.99545 DN
7D2 - DR 11.70 at 8MP (11.04 at 20MP), read noise 6.28073 DN

ISO 400 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00400NR0.CR2.HTM)
80D - DR 11.92 at 8MP (11.13 at 24MP), read noise 6.37543 DN

ISO 800 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI00800NR0.CR2.HTM)
80D - DR 11.09 at 8MP (10.29 at 24MP), read noise 11.3868 DN

ISO 1600 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI01600NR0.CR2.HTM)
80D - DR 10.25 at 8MP (9.45 at 24MP), read noise 20.3616 DN
7D2 - DR 10.41 at 8MP (9.75 at 20MP), read noise 15.4014 DN

ISO 3200 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI03200NR0.CR2.HTM)
80D - DR 9.41 at 8MP (8.62 at 24MP), read noise 36.395 DN

ISO 6400 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI06400NR0.CR2.HTM)
80D - DR 8.63 at 8MP (7.84 at 24MP), read noise 62.2559 DN
7D2 - DR 8.95 at 8MP (8.29 at 20MP), read noise 42.3891 DN

ISO 12800 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI12800NR0.CR2.HTM)
80D - DR 7.64 at 8MP (6.85 at 24MP), read noise 124.027 DN
7D2 - DR 8.00 at 8MP (7.34 at 20MP), read noise 82.8706 DN

ISO 16000 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI16000NR0.CR2.HTM)
80D - DR 7.33 at 8MP (6.54 at 24MP), read noise 153.517 DN

ISO 25600 (http://www.imaging-resource.com/PRODS/canon-80d/80DhSLI25600NR0.CR2.HTM)
80D - DR 6.74 at 8MP (5.95 at 24MP), read noise 231.801 DN

The new sensor uses a black level of DN (data number) 512 for ISO 100 and 200, but continues using 2048 for the remaining ISO levels. The saturation level seems to be the DN 16383 at all ISO levels, which is unusual. For example, the analyzed ISO 100 picture has 871 blue pixels with DN 16383, 1039 pixels with DN 16382, and 1660 plus 1632 green pixels (two green channels) with DN 16383; the image is overall underexposed and from DN 10000 to 16381/2, each DN only counts 0-2 pixels.

There are two masked areas in the 25.5MP image (6288x4056): one 264 pixels wide at the left and the other 34 pixels heigth at the top. So, the effective image area is 6024x4022 (one-two pixels in the frontier of both areas, showing visible artifacts, are not truly usable). The masked area does not receives light and its standard deviation can be measured to determine the read noise, and to infer from that the DR. The area used has been the left one, but the top measures almost the same noise levels (unlike with other sensors, where the top masked area has some oddities).



Mar 17, 2016 at 08:22 PM





  Previous versions of cgarcia's message #13468978 « Canon 80D dynamic range: good news! (quick test with RAW images) »