EX1R Noise

L

LeavingTheCandy

Well-known member
I seem to be having way too much noise on my EX1 footage. Gain is set to -3 and I'm recording to a KiPro.
What would be the main settings I should look at to check on this? I am NOT shooting with the color temp in the tungsten range. I keep it at appx 5500K and correct with gels, etc.

Thanks.

p.s. I shoot a lot of 5DMKII so that's my baseline comparison. Is this just a LOT less noisy?
 
I use an EX3 and a Ki Pro and I've never noticed any objectionable noise. However, I've never compared the footage to the 5D. I usually shoot a 0db as I've read that you lose dynamic range at -3. Also, I turn detail off and use Cinegamma1-operating under the premise of less stress on the sensors equals less noise. As far as color temp. goes, I use both 32k and 56k. If 56k is cleaner it's minimal. The key is to have enough light. I hope some of this helps.
 
picture profiles can play a big part in this. the black setting needs to be set around negative 4. That will black balance the camera. it being at zero or even people upping it to "get detail in the blacks" are just washing out their image, the black setting in the picture profile only adjusts what signal level absolute black will be. If you can, hook your ex1r up to a waveform monitor, close the lenshood so its completely black, and adjust the black setting until the waveform is at 0 IRE. for most cameras its between -3 and -5, so if you don't have a monitor to do this with I'd go with -4. anything above zero is limiting your dynamic range by adjusting the cameras output so no data can exist below whatever IRE you've upped the black point to. If you want to stretch out the blacks and get more information down there, that's what black gamma is for. But that will also add noise.

Also, a huge part of this in the picture profile detail settings. this is where a lot of the noise comes from. What I've settled on over the years are the settings form wolfgangs profile, plenty of sharpness but almost amazingly little noise compared to what I've been using. He devised it using a bunch of test equipment, his full profile is the most accurate one I've used if you're interested. Settings for detail are are:

Setting: on
level -30 (I've upped mine to around -20, no noticeable noise introduced)
frequency: -87
crispening: -80
H/V ratio: +25
white limiter: 0
black limiter: o
V DTL creation: NAM
knee apt level: -45

Also, if you have the low key saturation increased, this will also indefinitely add noise.

Also, try shooting at 0db and see what happens. running at -3db limits your whites and dynamic range to 100IRE, this has been tested. It completely cuts off all your superwhites. If you don't care about superwhites, then it's much better to use the cine2 profile instead of shooting at -3db, as the cine2 limits superwhites to 100ire, but in a much more forgiving and smooth rolloff fashion. -3db might actually even limit whites to 90IRE, I haven't had it on a waveform monitor in a while.

Also, STD3 is what is in wolfgangs profile, huge exposure range and haven't noticed any noise. What picture profile are you using? some of them that shift the color matrix huge levels (ie r-g -90 r-b -62) will also add noise
 
Fohdeesha would you share with us your PPs? I'm trying to find out which my favourites are and i recently ended in the ones with big shiftings (R-G: +75, G-R: -18, G-B: -32, B-R: -27, B-G: +13).
 
No problem, I've been using Wolfgangs profile that was posted somewhere around page 38 in the picture profile thread over on dvinfo. I've been through a lot and I'm definitely keeping this one, the neutrality and noise performance is great. It also taught me one thing, the cine curves are asthetically great at first glance, but for capturing the maximum amount of information and dynamic range for grading, std3 is really great. The auto knee feature of this camera works surprisingly well to stop things from blowing out that usually would. (EDIT: this is all lies! std curves are awful for dynamic range and highlight retention, the knee oversaturates highlights far to easy and ugly. The new defacto standard is cine1 when you have tons of light, if you are shooting in very low light where most of your information lies below the midtones or in the midtones with no highlights, use cine3)

heres a link to his post for the profile (its a bunch of images)

http://www.dvinfo.net/forum/sony-xd...2-picture-profile-recipes-38.html#post1540732

remember to have your black setting at -4 or so and black gamma at zero, unless you're trying to stretch the blacks like he is. doable but will add noise. He also has his low key sat set at +5, which i dont recommend as its asking for noise
 
Last edited:
What some awesome info...great info. Thanks.

I am not sure which profile I am using (a bastardization of a few, I think). I am gonna run to my office right now and get my camera to check on this stuff. (and play with it today).
Using it with the KiPro is a JOY compared to the 5D, but those 5D images are just great. I am hoping to ditch both cameras and get an F3 just as soon as I can get my mitts on one).
Imagine the look of that thing going to 10-bit ProRes! ;-)

I'll get back in a few with my profile settings report. I *think* I am using Wolfgang's -3 profile, now that I look at it.
 
My settings were identical (I was using Wolfgang's settings) including the +5 LOW KEY SAT. I changed that to zero and dropped my black to -5 (verified on my scope).
I'll do some tests. Will that +5 KEY SAT add THAT much noise?
Thanks.


UPDATE: Ran a new test. Yes, a LOT less noise with the +5 turned to zero. It now looks like I'd expect it to! Thanks! ;-)
 
Last edited:
Fohdeesha said:
It also taught me one thing, the cine curves are asthetically great at first glance, but for capturing the maximum amount of information and dynamic range for grading, std3 is really great. The auto knee feature of this camera works surprisingly well to stop things from blowing out that usually would.
Click to expand...

That's interesting to hear, I'll have to test it out.

I was always under the impression that the cine gammas provided a flatter image, and allowed for superwhites, while the standard gammas provided more contrast, and cut things off at 100%.
 
LeavingTheCandy said:
My settings were identical (I was using Wolfgang's settings) including the +5 LOW KEY SAT. I changed that to zero and dropped my black to -5 (verified on my scope).
I'll do some tests. Will that +5 KEY SAT add THAT much noise?
Thanks.


UPDATE: Ran a new test. Yes, a LOT less noise with the +5 turned to zero. It now looks like I'd expect it to! Thanks! ;-)
Click to expand...


sweet!
 
I'm going to have my EX1 calibrated to "neutral white" and then to "warm white" within the next two weeks. One of those big Gretag MacBeth (correction: DSC Labs Chromadumonde) charts will be used and it can't be overstated enough to get your individual camera(s) calibrated using one of these charts with a PROPERLY calibrated production monitor and vectorscope/waveform monitor. Each camera, even ones from the same model and serial # batch are slightly different in color rendition.

From my understand the GM chart aligns a certain way on the vectorscope when the colors are neutrally calibrated. Once that's done we'll tweak some of the color and black settings so that I can get a saturated/high contrast look for footage right out of the camera. Two profiles, one happy customer.
 
Last edited:
adamr316 said:
From my understand the GM chart aligns a certain way on the vectorscope when the colors are neutrally calibrated. Once that's done we'll tweak some of the color and black settings so that I can get a saturated/high contrast look for footage right out of the camera. Two profiles, one happy customer.
Click to expand...

You want a ChromaDuMonde chart, not a MacBeth. A Macbeth doesn't line up anything at all on a vectorscope. A DSC CrhommaDuMonde, or CamAlign chart does exactly what you are describing. And yes, calibrating to a chart is something that certainly should be done. In fact, for critical work, it should be done prior to each shoot in the prevailing light. But that seemingly only happens on pro shoots.
 
PerroneFord said:
You want a ChromaDuMonde chart, not a MacBeth. A Macbeth doesn't line up anything at all on a vectorscope. A DSC CrhommaDuMonde, or CamAlign chart does exactly what you are describing. And yes, calibrating to a chart is something that certainly should be done. In fact, for critical work, it should be done prior to each shoot in the prevailing light. But that seemingly only happens on pro shoots.
Click to expand...

Sorry I am getting my names all confused! I've been busy learning so much new stuff these past few months that I'm getting names and functions mixed up.

All I know is the chart is huge, has black felt in the center and cost a lot of money. That's the chip chart I'll be using, LOL.
 
PerroneFord said:
The ChromaDuMonde chart is quite expensive. But the CamAlign charts are not. And work quite well. I've used them in the past on 3-cam shoots so we didn't have to kill ourselves in post getting the footatge to line up.

Go here:

http://www.dsclabs.com/Mini-Chart.htm



Good info here:

http://www.dsclabs.com/DSC_Movies.htm
Click to expand...


Thanks for the link Perrone! My buddy already owns the chart and I'm finally going to take my EX1 over to him to calibrate. If I ever need to get my own chart I'll definitely keep CamAlign in mind.
 
You must log in or register to reply here.
Back
Top