Question for Canon 7D owners

[Tell Dickinson]

Quick question if any 7D owners have tried this or can do some tests...

Most of the 7D examples on BPN have shown acceptable noise at iso400 IMHO, the 7D has a respectable 10mp RAW setting which may be useful to get better quality (at the expense of mp) high iso pictures than when using the full 18mp, has anyone done any testing to see if noise is actually better at this setting or would just downsizing a 18mp image to 10mp look similar ?

Tell

[arash_hazeghi]

I would just downsample in my PC afterwards, if you shoot sRAW in camera you cannot recover the original 18mpixel detail if needed.

[Tell Dickinson]

if you shoot sRAW in camera you cannot recover the original 18mpixel detail if needed.

Hi arash, thanks for the reply, yes that is true but there may be times when someone would rather get a smaller (in mp) but cleaner image and until a test is made then we can only 'assume' quality/noise will be equal/similar :)

Tell

[Ed Cordes]

While this is a good question, no one outside the Canon engineers knows exactly how Canon achieves the smaller mRAW and sRAW files. Are they actually using less pixels on the sensor, or downsizing the full RAW image. My guess is downsizing the original. So, with this in mind I agree with Arash. Shoot in Full RAW then downsize yourself.

[Simon Bennett]

Tell, 7D noise is very fine grained and seems quite easy to clean up - here's an example: http://www.birdphotographers.net/for...ad.php?t=47188

[Tell Dickinson]

Hi Simon, thanks for that, I realise that 7D noise at 'normal' iso's are ok but I am just wondering if high iso images (say 32000) are actually better at 10mp than 18mp as opposed to down sampling, its another tool to have in your bag if you know what actually happens rather than just assume it doesnt make any difference :)

Tell

[Roger Clark]

Tell,
Although I have not tested it, I would expect the in-camera 10mp to produce a better result than standard down sampling. Down sampling algorithms, e.g. by cubic spline will not do any averaging. You would need to average pixels to get any benefit in your post processing. (Or blur then cubic spline down sample). Try it and see!

Roger

[Tell Dickinson]

Tell,
Although I have not tested it, I would expect the in-camera 10mp to produce a better result than standard down sampling. Down sampling algorithms, e.g. by cubic spline will not do any averaging. You would need to average pixels to get any benefit in your post processing. (Or blur then cubic spline down sample). Try it and see!

Roger

Thanks Roger, your thoughts are the same as mine in that (hopefully) the 10mp image uses effectively larger pixels (by averaging) than mere down sizing so you should get less noise. I have not got a 7D to test with which is why I was asking if anybody had already done this, I know I could try on a 50D but wanted really to see what difference was like on a 7D as I am thinking of upgrading soon and it would be nice to know, not essential but nice :)

Thanks again Roger.

Tell

[arash_hazeghi]

Hi arash, thanks for the reply, yes that is true but there may be times when someone would rather get a smaller (in mp) but cleaner image and until a test is made then we can only 'assume' quality/noise will be equal/similar :)

Tell

Tell, you can do averaging in post later if you want. Also Lightroom sometimes has problems renderings the sRAW files correctly. DPP is fine.

[john crookes]

Even though you are downsizing the resouloution of the image to 10mp

the actual cells on the sensor would be the same size as the ones used for the 18mp file i do not see how the noise wuld be any better at the lower resoulution and you would be giving up way to much information that you would not be able to regain later

[Tell Dickinson]

the actual cells on the sensor would be the same size as the ones used for the 18mp file i do not see how the noise wuld be any better at the lower resoulution

"effectively larger pixels" by averaging the signals from more than 1 pixel instead of just reading each individual pixel, but that is EXACTLY why I am asking because until it is tested everyone will give their educated guesses without any real life data to back them up.

Tell

[john crookes]

from all my reading to do that would entail at least a 1/4 size resoulution change it is called pixel binning and canon says that is not what is going on
Nikon uses a lossless compression in their raw files but from my testing it does not change the final file size or affect noise

here is a link to a discussion

http://forums.dpreview.com/forums/re...ssage=32885962

[Don Saunders]

Check out my recent post of ISO tests all the way to 3200!