HVR-A1/HC1 low light noise and Mike Crash DNR

Laurence wrote on 3/17/2006, 9:40 PM
I have an HVR-A1 which produces an incredible picture in good light but in low light with gain gives an image absolutely crawling with noise. Anyway, I have found that the free Mike Crash "dynamic noise reduction" absolutely works miracles in getting rid of this noise. IMO this plugin can rescue footage you otherwise wouldn't be able to use. Anyone with an A1 or HC1 who doesn't have this plugin really needs to get it.

Comments

Laurence wrote on 3/17/2006, 9:48 PM
Here is where to find the dynamic noise reduction plugin:


http://mikecrash.wz.cz/vegas/vegas.htm
Spot|DSE wrote on 3/18/2006, 9:55 AM
Be sure you ALWAYS RAM render anything using this tool, because it looks at forward frames, and viewing a single frame, you can't tell what it's really doing. If you apply too much DNR, you end up with ghosting, so be cautious.
It is indeed, a great tool, however. But like any other temporal tool, it requires a good preview before rendering.
johnmeyer wrote on 3/18/2006, 11:44 AM
I want to second Spot's caution. Virtually everyone that uses denoising tools, makes the mistake of applying too much correction. You get so giddy at seeing all that horrible grainy noise magically disappear that you don't at first notice all the bad things that started happening to your video. My usual advice to people just getting started is to find a setting that seems about right and then use 1/2 to 2/3 that amount.

I regularly use some unusually advanced noise reduction, using a tool called AVISynth. I am in the midst, right this minute, of noise-reducing some old 1930s cartoons captured on VHS videotape. Given the flat 2D look, I thought I would be able to really crank up the temporal noise reduction. Turns out, I had to use less rather than more because I ended up with some really nasty contouring on what should have been large areas that were filled in with exactly the same shade of gray (these are B&W). Didn't notice it until I created a test DVD and looked at it in the monitor. Here's the AVISynth script I'm using. I edit on the Vegas timeline, then use the Debugmode frameserver to serve out of Vegas into this AVISynth script. I then serve this into VirtualDub, and immediately serve out of that (using direct stream copy) into the MainConcept external encoder (which for some reason won't read the AVISynth script, which is why I need what is otherwise a pointless step of serving into and then immediately out of VirtualDub).

# Script to IVTC a movie.
# Note: must serve into VD and then direct stream serve into MC
# For this script, serve out of Vegas using YUYV

#-----------------------------
#Telecide and Decimate are part of the Decomb plugin

loadPlugin("c:\Program Files\AviSynth 2.5\plugins\Decomb\Decomb.dll")
loadPlugin("c:\Program Files\AviSynth 2.5\plugins\CNR\Cnr2.dll")

AVISource("E:\Frameserver.avi")

AssumeBFF()
converttoYV12(interlaced=false)
Telecide(show=false,guide=1,post=2,vthresh=45)
Decimate(cycle=5)

Cnr2("oxx",8,16,191,100,255,32,255,false)
FluxsmoothT(5)

AssumeFPS(23.976, true)

ConvertToRGB32(interlaced=false) #Needed to get correct color when serving into MC.
Levels(16, 1, 235, 0, 255, coring=false)
As you can see, I am also doing inverse telecine (IVTC) to recover the original 24 fps progressive frames prior to encoding. This is virtually mandatory if you want to get a clean DVD encode from material that was previously telecined. The difference is night and day.
Laurence wrote on 3/18/2006, 1:44 PM
I've never been that much of a fan of video noise reduction, but with the Sony HDV CMOS cams it really works well. I think to key reasons for this are:

1, The sensitivity is calibrated for SD, so with HDV there is a lot less NR applied for a given setting.

2, Since most of us are shooting in HDV but outputting SD DVDs, the usual loss of resolution is not that big a deal.

3, For some reason this plugin works really well with the CMOS video noise.