View Full Version : Noisy EX3 shots
Peter Brinkman February 26th, 2014, 03:43 PM I lately notice a lot of noise in the dark parts of my shots. As you can see in enclosed still.
This is shot with the EX3 gain on 0 - 1080i - shutter speed on 60 - using the pp Doug prefers.
Normally I shoot on -3db and shutter speed off but al so then there is noise visual.
Any ideas?
TIA
Peter Brinkman February 26th, 2014, 03:48 PM Here is the still
Shaun Roemich February 26th, 2014, 03:56 PM If you are talking about the noise in the blue seat back, that looks pretty typical of an EX3 in a mid-to-low light scenario to me.
The 35mbps codec is a reasonably noisy one, especially when compared to the 50mbps mode of the PMW200/300 cameras.
Peter Brinkman February 26th, 2014, 04:04 PM Hi,
Yeah it's shot in a low light situation. But I had the idea that if I could hold the gain on 0 with iris on 1.9 it would be ok. No picture profile can change this I think.
Attached another still.
Chris Medico February 26th, 2014, 05:03 PM Did you by chance shoot this with CINE4? That gamma curve lifts the dark areas and will make the noise a lot more noticable.
To minimize the look of the noise in the blacks it is best to use a gamma curve that lowers the blacks such as CINE1.
Peter Brinkman February 27th, 2014, 02:19 AM Yes it's on CINE4. It is the recommended setting for overall shooting in the Field Guide I have. But I'll have a look at a setting which will handle low light better.
Thanks.
Chris Medico February 27th, 2014, 07:53 AM CINE4 is better to boost the bottom end which leads to a better image in low light. Noise will be more apparent using that curve. Use it only when necessary.
Jacques Mersereau February 27th, 2014, 04:17 PM A couple of things to note from what I've noticed with the EX3.
The EXCAM 35mbps codec is the issue with the blacks even at -3db.
The camera sensor exhibits quite a bit of noise even when we shoot uncompressed (green screen) via Kona3 and a raid.
But, if you can record ProRes to an external recorder, the blacks look much cleaner.
If you have the wherewithal, neat video does a pretty good job of cleaning up that noise.
Chris has some good suggestions, and perhaps some CC might also be sufficient.
As always, WMMV.
Doug Jensen February 27th, 2014, 07:23 PM The difference you're seeing in picture quality when using an external 10-bit capable recorder is mostly due to 4:2:2 vs. 4:2:0 and 8-bit vs. 10 bit -- and not really so much the bit rate. It is nearly impossible to see any visual difference between 35Mbps, 50Mps or even 100 Mbps if all other things are equal.
So, the important difference between internal recording on a EX3 and a PMW-200 is really the 4:2:0 vs. 4:2:2 -- the higher 50 Mpbs bit rate is just icing on the cake but doesn't make a big difference.
Shaun Roemich February 27th, 2014, 07:35 PM So, the importance difference between internal recording on a EX3 and a PMW-200 is really the 4:2:0 vs. 4:2:2 -- the higher 50 Mpbs bit rate is just icing on the cake but doesn't make a big difference.
To be fair, Doug, the 50mbps FACILITATES the extra information PROVIDED by the 4:2:2.
Doug Jensen February 27th, 2014, 07:37 PM No, they are two independent settings. I can easily record 4:2:2 at any bit rate I want with the appropriate external recorder or a software encoder on a computer. So, sorry, they have nothing to do with each other.
Shaun Roemich February 27th, 2014, 07:40 PM They DO with the XDCam codec set.
35 mbps is 4:2:0 and 50 mbps is 4:2:2.
I'm SPECIFICALLY referencing your SECOND paragraph in your above statement regarding INTERNAL recording.
So, the important difference between internal recording on a EX3 and a PMW-200 is really the 4:2:0 vs. 4:2:2 -- the higher 50 Mpbs bit rate is just icing on the cake but doesn't make a big difference.
Doug Jensen February 27th, 2014, 07:54 PM The 35mbps codec is a reasonably noisy one, especially when compared to the 50mbps mode of the PMW200/300 cameras.
I am referencing your comments above which is completely wrong. There is no difference in noise between 35Mpbs and 50Mbps when everything else is equal.
Doug Jensen February 27th, 2014, 07:55 PM They DO with the XDCam codec set.
35 mbps is 4:2:0 and 50 mbps is 4:2:2.
I'm SPECIFICALLY referencing your SECOND paragraph in your above statement regarding INTERNAL recording.
Yes, and I stand by that statement 100%. What is your issue with it?
Shaun Roemich February 27th, 2014, 08:02 PM There is no way to record 4:2:2 XDCam at 35mbps.
Therefore the statement that "the higher 50 Mpbs bit rate is just icing on the cake but doesn't make a big difference." certainly appears to me to be double-talk and erroneous. One cannot record 35mbps 4:2:2 XDCam and to GET 4:2:2 XDCam one needs to record 50mbps, the additional 15mbps being allocated to the additional colour information unless I am seriously misinformed.
ADDENDUM: not QUITE as simple as that - the 15mbps is very likely allocated not JUST to the extra colour but I would argue the additional colour depth certainly makes up the lion's share...
Doug Jensen February 27th, 2014, 08:12 PM There are ways of recording 35mbps 422 with external recorders. But that was not my point anyway. My point is that the 35Mpbs bit rate is NOT what it is important for better picture quality -- what is important is the 422 color resolution. If the 50Mpbs comes along with it, fine, but that is not the important part of the codec.
Why don't you do this: Get a NanoFlash and record XDCAM 4:2:0 at 35Mpbs and 100 Mbps and see if you see an ounce of difference in noise between them. If not, then clearly the bit rate is not the important ingredient of the codec's recipe. I know, because I have actually done the test.
Like I said, it is the 422 vs. 420 that is important, and if that is what you meant to say in your first post then you should have chosen your words more carefully to be more accurate in what you were trying to say.
If you don't understand that, then I give up. In fact, I give up anyway because I have better things to do that argue about dumb stuff like this. You are more than welcome to have the last word and refute everything I have said. Knock yourself out.
Jacques Mersereau February 28th, 2014, 10:23 AM Easy Brothers :-D
Unless I have this wrong,
4:2:0 = noisy blacks at whatever bit rate (because of the hideous noise in the blue channel?)
4:2:2 at 50mbps (or higher) make the blacks look much better - probably because the blue channel is much less noisy.
EX3 is a great camera for what it does, but there are things that are not quite up to snuff.
The sensor is noisy (when compared to say the F3)
Many times I have noticed that there must be IR contamination, because on certain fabrics- say black pants, they look brown/maroon.
(and no, I don't want to have to use an IR filter and lose a stop or two)
The zoom on the stock lens does not do a super slow crawl without hiccups when using external zoom controller.
As always, YMMV.
Doug Jensen February 28th, 2014, 11:49 AM Jacques, I am in complete agreement on everything in your post.
Alister Chapman March 2nd, 2014, 12:28 PM Bit rate does effect noise with high compression codecs as the lower the bit rate the more quantisation noise and macro blocking you get. With the same codec and higher bit rates there is less compression noise, typically seen as mosquito noise. Macro blocking and quantisation noise often manifests itself as banding across near even surfaces or the sky etc.
The 422 recordings of the PMW-200 need more data than the 420 recordings, so much of the data difference between 35Mb/s and 50Mb/s goes purely on the extra data, but finer sampling of the chroma will result in less coarse noise which tends to be less noticeable or objectionable than coarser noise.
If the bit rate wasn't important then we would all be shooting 25Mb/s 422, but the bit rate does matter with a higher bit rate generally resulting in fewer artefacts including quantisation noise.
I did many tests with the Convergent Design NanoFlash when that came out comparing different bit rates and there was a useful noise improvement when going from 50Mb/s to 100Mb/s. The viewer would may not always see this difference, but if you want to grade your material then the reduced quantisation noise means less likelihood of seeing any banding in your footage and the ability to grade the material with less degradation. Compression noise is a bigger cause of banding in footage than "only" 8 bit encoding.
Garrett Low March 2nd, 2014, 01:05 PM Alister, thanks for that explanation. I have experienced exactly what you have said in your post. which is why my usual recordings on my EX3 are done at 100 Mb/s on my nanoFlash when using Long-GOP (of course I also record to the internal cards). There is a notice a difference when comparing the internal 35 Mb/s to the nanoFlash recordings at 100 Mb/s. I can't understand how people don't see it. It might have something to do with the circuitry of the nanoFlash vs. the EX3's but there is a difference.
Jacques Mersereau March 2nd, 2014, 04:03 PM Friends don't let friends try to CC 35mbps.
|
|