HDR AGC Filter - lighten dark footage

[gb_direwolf]

Hi All

I am having a go at using AVISynth and the HDR AGC filter to lighten some footage that is too dark. This has been a steep learning curve for me to get this far so I want to double check to make sure I'm on the right track (I was going to ask the filter author over at the doom forums but you have to be registered for 5 days before you can post - WTF??)

Anywhoo

This is the script I used

AVISource("G:\Video editing\capture\dv-cap.03-12-02_17-50.00.avi")
ConvertToYV12(interlaced=true)
hdragc(max_gain =3.0, min_gain=1.0, coef_sat=2.5)

And this is my results


Is there anything more I can do to perhaps smooth out the grainy look of the lightened footage?

[guns1inger]

You could try one of the denoising filters afterwards. I like both PeachSmoother and Convolution3D. Both are worth a look to see if they help. But keep it light or you will have a nice plastic feel to it.

[gb_direwolf]

Thanks

This helped clean it up a bit

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3DYV12.dll") 
AVISource("G:\Video editing\capture\dv-cap.03-12-02_17-50.00.avi")
ConvertToYV12(interlaced=true)
HDRAGC(coef_sat=2.5,max_gain =3.0,min_gain=1.0,corrector=1.5)
SeparateFields()
odd=SelectOdd.Convolution3D (0, 32, 128, 16, 64, 10, 0) 
evn=SelectEven.Convolution3D (0, 32, 128, 16, 64, 10, 0) 
Interleave(evn,odd)
Weave()
DoubleWeave.SelectOdd()

[SerbianBoss]

HDRAGC need to be between separate fields/weave when applaying on interlaced material

[gb_direwolf]

What, like this?

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3DYV12.dll")
AVISource("G:\Video editing\capture\dv-cap.03-12-02_17-50.00.avi")
ConvertToYV12(interlaced=true)
SeparateFields()
odd=SelectOdd.Convolution3D (0, 32, 128, 16, 64, 10, 0)
evn=SelectEven.Convolution3D (0, 32, 128, 16, 64, 10, 0)
Interleave(evn,odd)
HDRAGC(coef_sat=2.5,max_gain =3.0,min_gain=1.0,corrector=1.5)
Weave()
DoubleWeave.SelectOdd()

[manono]

No, not like that.

If you really want to do it right:

LoadPlugin("C:\Path\To\LeakKernelDeint.dll")
LoadPlugin("C:\Path\To\Convolution3D.dll")
MPEG2Source("C\Path\To\Movie.d2v")
ConvertToYV12(interlaced=true)
LeakKernelBob(Order=1)#if TFF,Order=0 if BFF
HDRAGC(coef_sat=2.5,max_gain =3.0,min_gain=1.0,corrector=1.5)
Convolution3D (0, 32, 128, 16, 64, 10, 0)
SeparateFields()
SelectEvery(4,0,3)#if TFF, SelectEvery(4,1,2) if BFF
Weave()

And encode for TFF. Or, if it matters to you, you can make it for BFF by changing some things.

[gb_direwolf]

Sorry, what is TFF and BFF?

[guns1inger]

Top Field First
Bottom Field First

Describes the field order of interlaced video. DV is Bottom (or Lower) Field First.

[gb_direwolf]

ahhhh ok

So, as my video is a DV capture, it should look like this

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\LeakKernelDeint.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\Convolution3D.dll")
AVISource("G:\Video editing\capture\dv-cap.03-12-02_17-50.00.avi") 
ConvertToYV12(interlaced=true)
LeakKernelBob(Order=0)#if TFF,Order=0 if BFF
HDRAGC(coef_sat=2.5,max_gain =3.0,min_gain=1.0,corrector=1.5)
Convolution3D (0, 32, 128, 16, 64, 10, 0)
SeparateFields()
SelectEvery(4,1,2)#if TFF, SelectEvery(4,1,2) if BFF
Weave()

[manono]

That's the way I'd do it. I don't know the output format, so there's a chance you'll have to change the colorspace again. If it's for XviD or HCEnc, you're OK. The Doom9 guys might suggest adding in some AssumeTFFs, as AviSynth has a disconcerting habit of switching back to BFF when you least expect it, but I think you're OK with those filters. If you're not sure, add an Info() to the bottom of the script temporarily to check. Also, before encoding, open the script in VDubMod, find a place with movement , advance a frame at a time, and make sure it plays smoothly.

And if SerbianBoss comes back, he prefers to encode for BFF, and might suggest a slight change.

And I didn't notice that you had DV in the name. I didn't even notice it was an AVI. So my source line was incorrect. But between yourself and guns1inger you got it right.

[gb_direwolf]

Cool. Thanks for all the help guys

[SerbianBoss]

I'll be back

Maybe add AssumeBFF() before leakkernelbob and AssumeBFF() after Convolution 3d.

For BFF material i am using SelectEvery(4,0,3)