Archived

This topic is now archived and is closed to further replies.

Sony RX100: Getting the best video out of it...


153 posts in this topic

Posted

It happened to be on my desk at the moment, so I've done a quick test. Stills from 1080/60p videos. I think they're as near as identical! Or if there is a difference I'm struggling to see it. Perhaps the reds seem a tiny bit darker/saturated in the Adobe ones, but I am looking for a difference, which can skew judgement. What do you think?

 

sRGB

SRGB_zpsb8890cab.png

 

Adobe 1998

ADOBE_zpsd9a164c0.png

 

EDIT:

 

Tried it with my hand for more complex tones. Pretty much the same again! (It's a slightly different angle hence the darker index finger, the greatest of all picture profiles: LIGHTING! ;) )

 

in sRGB:

2sRGBHAND_zps70b5fef9.png

 

in Adobe 1998:

2ADOBEHAND_zpsf772315a.png

adobeRGB looks less saturated/flatter from here.  

Share this post


Link to post
Share on other sites

Posted

Yeah adobeRGB does seem to be slightly different (less saturated & flatter).

 

Looking at the reflection in the mug & looking through the bottle, it also seems to give the impression of producing slightly more detail.

 

But with the skin tones it appears to smooth out the blemishes, thus making it more uniform - but whether the detail comes back when you cc in post...?  

 

I could be wrong, but could using adobeRGB produce better detail in the shadows & less in the highlights?

 

gotta try this....

Share this post


Link to post
Share on other sites

Posted

I suppose the test will be the classic "can you make one look like the other". It looks like a contrast shift to me more than a hugely different look. Now these differences have been mentioned I can see a subtle difference. The sRGB looks marginally flatter...

Share this post


Link to post
Share on other sites

Posted

If someone has a chart and an RX100 they could do a scientific test, I tend to do ones that look like the kind of thing I'll shoot! From that evidence here I think I'll stick with sRGB as the reds seems more squashed in Adobe. Looking at them as full 1080p (Photobucket downscaled them) there's even less difference. I'm thinking most of the change may have come from the Photobucket re-encoding.

Share this post


Link to post
Share on other sites

Posted

Judging from these images I'd say the Color Space has no effect on the video mode :)

I put both images one on top of the other and I don't think there's any difference. There's a slight angle change which can make a big difference on reflective surfaces such as the mug, but overall I don't see any change in color at all, not the kind of difference you'd see between sRGB and AdobeRGB at least.

Might be worth shooting some stills too to see if the difference is more apparent, and then compare it to video footage of the same thing to find out which color space the camera is using on video mode.

Share this post


Link to post
Share on other sites

Posted

Good idea. I feel the difference is too subtle to be the difference between sRGB and Adobe1998 though, if you look at the kind of difference it should make.
 
My testing wasn't that consistent (it was five minutes at my desk between edits) so I think you can attribute the difference to that, realistically. It's probably just Rec709 space all round.
 
One chart here shows it's very subtle though, maybe it would explain the difference in the reds? Still, for my purposes I've concluded it's not worth worrying about, so I'll leave it on sRGB!
 
chromacity-all.jpghttp://www.fxguide.com/wp-content/uploads/2011/02/chromacity-all.jpg]

Share this post


Link to post
Share on other sites

Posted

on the subject of DRO. it appears to work best at iso200 from some experimentation I have done.


assuming you are exposing to avoid clipping of the sky, the shadow noise on darker areas boosted by DRO seems to be more prominent at iso100 than the shadow noise caused by DRO when iso is set to 200 with the lens aperture adjusted to match exposure of the sky.

This makes me think that base ISO of nex5n is actually 200 (as it was with the original nex-5 and nex-3), but is having some type of post exposure compensation to allow for the option of iso100. The processing is naturally going to cause some type of image degradation. This is likely the case with the rx100 too.

Maybe some one can look into this. If true it might be worth avoiding iso100 and starting at iso200 to avoid banding and other nasties. It is known that this is a main cause of banding/colour issues with the Canon 1DC when used at ISO lower than the actual Base iso of the sensor. Being that the previous nex models were base iso of 200 it seems funny sony would downgrade on the newer model.. rather than just applying some processing to offer a faux lower iso of 100.

Thoughts??

Share this post


Link to post
Share on other sites

Posted

Haven't tested it myself, most people seem to agree it's 125 though, also that's what's mentioned in the manual.
Some people claim that the highest DR at 200 ISO only happen when shooting JPGs with noise reduction on, which would make the improvement a post effect kind of thing, and not a real improvement, was this the case with your tests?

Share this post


Link to post
Share on other sites

Posted

I have NR set to minimum and am talking in movie mode terms.

 

 

I am just adjusting my aperture to compensate for the difference in ISO.  It is just a lot less grainy in the shadows/darker areas (the areas boosted by DRO) when set to iso200.  Iso100 is significantly more grainy than iso200 in the DRO boosted areas.    

Share this post


Link to post
Share on other sites

Posted

Interesting!

 

I didn't realize that you can set NR in video mode. Where is that setting?

Share this post


Link to post
Share on other sites

Posted

Interesting!

 

I didn't realize that you can set NR in video mode. Where is that setting?

 

Menu, settings, high ISO NR, Low.  I'm unsure what is determined as 'high iso' but it seems to affect the noise levels at all iso's

Share this post


Link to post
Share on other sites

Posted

I saw that, but I can't seem to set it when I am in video mode. I assume it's because I am not shooting in jpg. I would assume that the NR doesn't happen in video mode based on that, but I could be wrong. I would have to do some tests. Did you notice that it was doign NR in movie mode?

Share this post


Link to post
Share on other sites

Posted

I am just adjusting my aperture to compensate for the difference in ISO.  It is just a lot less grainy in the shadows/darker areas (the areas boosted by DRO) when set to iso200.  Iso100 is significantly more grainy than iso200 in the DRO boosted areas.    

 

Did you try doing the same test with DRO off? Is it still noisier at 100 ISO?

 

Btw, 100 ISO (and 80 ISO) is being pushed in post, 125 ISO should be the base one, and therefore offer cleaner results, have you tried that one instead?

Share this post


Link to post
Share on other sites

Posted

Did you try doing the same test with DRO off? Is it still noisier at 100 ISO?

 

Btw, 100 ISO (and 80 ISO) is being pushed in post, 125 ISO should be the base one, and therefore offer cleaner results, have you tried that one instead?

 

without dro there doesnt seem to be a difference in noise between 100 and 200.  it's only when you go to iso100 with dro on that noise seems more prominent than at iso200.  PS.  this is on the nex5n, not rx100.  but since they share so many similarities with processing, etc I imagine these results will be similar on the rx100

Share this post


Link to post
Share on other sites

Posted

They have different sensors though, so it's very possible that their base ISOs are different. In the RX100 ISOs 80 and 100 are not proper ISOs, they're even marked differently. It's definitely worth testing though.

Share this post


Link to post
Share on other sites

Posted

Here's an ISO test.

These are all the ISO modes available during video mode.

125 is the cleanest, it's been mentioned in a few places that it's this camera's base ISO and it does look that way.

 

The only surprise here is that 400 ISO is actually less noisy than 200 ISO.

 

But overall, unless you push it really hard, the noise is barely visible before 3200 ISO.

 

ISO%20NOISE.jpg

Share this post


Link to post
Share on other sites

Posted

I'd say 400 is native probably. The fact that it has so little blue-channel noise (the biggest problem with current Sony sensors in my experience) is great to know, cheers for this test!

Share this post


Link to post
Share on other sites

Posted

I'd say 400 is native probably. The fact that it has so little blue-channel noise (the biggest problem with current Sony sensors in my experience) is great to know, cheers for this test!

 

I agree.  400 seems the best noise vs gain ratio.

Share this post


Link to post
Share on other sites

Posted

In that sense you could treat the RX100 like 400ISO Super 16 film, leave the ISO fixed for the most part and light to the sensitivity

 

The sensor is about that size!

Share this post


Link to post
Share on other sites

Posted

Here's a comparison between 125 and 400 ISO.

I pushed the 125 version one stop further than the 400 one, so they're equally bright when desaturated.

It's definitely a different noise structure, but I seriously don't know what to make of it.

125 does seem cleaner, and keep in mind that it's been pushed further than the 400 ISO side.

 

Looking at how good the 400 ISO version looks though, I wouldn't be surprised if it is indeed the camera's base ISO, but it does seem a bit high of an ISO for them to go with, on the other hand it could explain why this little sensor can do so well in low light. Also, if 400 ISO is in fact its base ISO, then they're doing a great job when pushing it to 125, much better than when pushing it to 200, which is somehow odd.

 

This article (and most others I found) also claims that the base ISO is 125.

 

http://www.imaging-resource.com/PRODS/sony-rx100/sony-rx100A.HTM

 

ISO%20NOISE%20%28125%20vs%20400%29.jpg

Share this post


Link to post
Share on other sites

Posted

Red and blue noise below 400, red noise only at 400, and red blue and green noise above 400. What a puzzler!

Share this post


Link to post
Share on other sites

Posted

these tests have prompted me to do some lens cap tests of the nex5n, which seems to exhibit the same varied noise colour depending on iso setting..  I'm uploading now

Share this post


Link to post
Share on other sites

Posted

how do I embed a video?

Share this post


Link to post
Share on other sites