[SOLVED] Editor to increase brightness of half of screen only?

[yetanotherlogin]

Even after massaging the file using the ffmpeg command above, still no trace of the file in the timeline

[poisondeathray]

really ? it worked for me. Try deleting the old index first then re-import (the .lwi file)

[jagabo]

Using the two image + mask technique I gave earlier:

[yetanotherlogin]

really ? it worked for me. Try deleting the old index first then re-import (the .lwi file)

Yup. Delete the two LWI files, dropped the fixed MPEG file: Still nothing in the timeline, save the vertical thin blue line mentioned before.

Using the two image + mask technique I gave earlier:

Beautiful. That's exactly what I need to do.

---
Edit: Ran the following command, dropped the file onto the Layer 1 line in the timeline, and lo and behold!

ffmpeg -i full.mpeg -c:v copy -c:a copy -f vob full.mp4

Are we positive AviUtl actually supports MPEG files?

[poisondeathray]

---
Edit: Ran the following command, dropped the file onto the Layer 1 line in the timeline, and lo and behold!

ffmpeg -i full.mpeg -c:v copy -c:a copy -f vob full.mp4

Are we positive AviUtl actually supports MPEG files?

Yes, positive. Interesting that it worked as MPEG2 in a MP4 container for you ...WTF ?!

This is in 1-6[xp] on a Win8 OS - I tend to use that one, whenever there is 5.1 audio. But 1-10[win7] also opens it up , but with wrong framecount (it's misreading interlaced content). 1-6 does it correctly with the correct 0-500 frames = 501 total, interlaced




But it sounds like you're leaning towards avisynth...

This is the variation using the HDRAGC mentioned earlier. It was bob deinterlaced, and 2nd instance of HDRAGC is used on the highlights by using Invert()

Anyways, there are many ways you can approach this - there are many filters within aviutl or avisynth you can use to get similar results. Curves mentioned earlier would also work in vdub, but you need to convert it to full range RGB first, otherwise you clip the overbrights

[poisondeathray]

And do you have a copy of the speaker's slides ? Or could you ask for them? Because that would be a way to clean up the projection/display you could composite them on with a corner pin to match the perspective

[yetanotherlogin]

After finally figuring out how to display the right filters and using the settings you provided (Luminance option), I'm currently saving the file as an AVI file. I'll report back once it's done.

Yes, I intend to insert the slides into the video but I still wanted to somehow brighten up the video so that the speaker isn't barely-visible the whole time.

[jagabo]

Yes, I intend to insert the slides into the video but I still wanted to somehow brighten up the video so that the speaker isn't barely-visible the whole time.

In that case you can brighten the entire frame and let the projection screen blow out. Then you just overlay new slides over it.

[poisondeathray]

Because you have full range values, it might be a good idea to use the adv. color correction checkbox (filter+ => adv. color correction) . It needs to be checkmarked in the checkbox to be enabled. The reason I'm suggesting it is there is a PC=>TV range conversion checkbox at the bottom. That will ensure nothing is clipped before you start to fix it. Some of the filters you use might work in RGB, and you will get clipping if you use one of them if you don't normalize values in YUV first

And wouldn't it be better to render it once (ie. do everything in one go, fix the shadow brightness, and insert the slides) ? In that case you don't care about overbrights because the inserted slides will "cover" that area up anyways

[poisondeathray]

BTW "corner pin" effect in aviutl to distort an image is under animation=> simple deformation

You just drag the corners to match up where you want them so the perspective matches. You would insert your "slide" clips on track 3, because higher numbered tracks "cover up" lower numbered tracks (reverse to what you'd be used to in something like photoshop or an image editor). So you 'd still brighten the shadows in whatever way that you want, you can composite the slide on track 3

If it's all static (no camera movement) once you position the 1st slide you can just copy & paste the effect to the other slides (because the location will be the same) . (Those dotted lines in the preview won't be there when you render it, it's just in the preview)

[yetanotherlogin]

I'll insert full width slides, because displaying them on the right half would be too small.

BTW "corner pin" effect in aviutl to distort an image is under animation=> simple deformation

Amazing Thanks much for the tip.

[poisondeathray]

But the point about rendering multiple times still stands - wouldn't it be better to insert the full slides into the project , do it all in one go ?

[racer-x]

Yeah, I forgot about "Simple Deformation"...........

[yetanotherlogin]

Yes, I'd rather brighten up and insert the slides in one go, to only render the output once.

Now that it appears to be working, I have a couple of questions:

1. In AviUtl, is it better to use the clipping/cropping option or the luminance option?
2. In case I use the luminance option, do I still need to copy the video track, since the original track won't be worked one and won't show through?
3. To see how the output differs, I'd like to perform the same task with AviSynth. Would someone have a working script that I can call through VirtualDub?

Thank you.

---
Edit:

1. In AviUtl, when saving the file as AVI with default settings, its ETA is 12 hours, and with just four seconds of video, the file is already 60MB

2. Using VirtualDub 1.10.14: In Filters, no trace of a "Gradation Curve" filter. It can be found here:
http://members.chello.at/nagiller/vdub/downloads.html

Just unzip, and put the VDF file into VirtualDub's \plugins sub-folder.

NO GO : If I use "Full Processing Mode", the file is huge; If I use "Normal recompress", the Filter section is greyed out, and the output file has audio but no video (although VLC shows that the file has two tracks.

Just choose an option in the Video > Compression menu, and the file should be much smaller. Using the DivX 5.0.5 Codec, this 2h26m file is about 1,300MB. It can probably be trimmed out more with the right settings in ffmpeg.

3. I also tried AviSynth, but in both cases, it complains with "Avisynth open failure: Convert: Cannot convert if width isn't mod2!"

ImageSource("c:\full.mpeg")
ConvertToYV12()
HDRAGC(coef_gain=1.0, coef_sat=0.75)
HDRAGC(coef_gain=0.2, coef_sat=0.75)

alpha=ImageSource("c:\full.mpeg", start=0, end=23, fps=23.976).ConvertToYV12(matrix="PC.601")
ImageSource("c:\full.mpeg", start=0, end=23, fps=23.976).ConvertToYV12()

darks=ColorYUV(gain_y=600, off_y=-30) # a lighted version of the video
brights=ColorYUV(cont_y=400, off_y=-170) # a darkened version of the video

Overlay(darks, brights, mask=alpha) # blend the two together

[jagabo]

Image source is only for images (BMP, JPG, PNG, etc.). The best way to access MPG video is to build an index with DgIndex, then open the d2v file that creates in AviSynth with Mpeg2Source("filename.d2v"). You'll need to download and install the DgMpegDec package for AviSynth.

You'll need to make an alpha mask manually and open it with ImageSource().

A simple alternative using mostly a gamma adjustment:

Mpeg2Source("full.sample.d2v") 
ColorYUV(cont_y=-20, gamma_y=200, off_y=-25)

[poisondeathray]

In AviUtl, is it better to use the clipping/cropping option or the luminance option?

It depends on the project. If the entire video is like this shot I wouldn't use the clipping(crop) option. You can try both and see what you like. There are probably a few other variations you could use in aviutl too

In case I use the luminance option, do I still need to copy the video track, since the original track won't be worked one and won't show through?

Yes I would. You can also use other filters, enable the adv. color correction - the point of the earlier example was only to show the difference between clipping(cropping) and using a luminance key method to selectively apply filters . If you can get any mask based method working (either with resolve, avisynth, aviutl or any program), it looks more elegant than a 50/50 split IMO . That's just the basic approach of using mask to selectively filter which can be done in almost any video editor

1. In AviUtl, when saving the file as AVI with default settings, its ETA is 12 hours, and with just four seconds of video, the file is already 60MB

You need to use a compression format. If you want an AVI export format you can access any VFW codec that you have installed (basically anyone that shows up in vdub's compression list)

But the adv x264 export is probably the "best" and most popular. You've used libx264 in ffmpeg right? It's basically the same thing, but in GUI format.

[yetanotherlogin]

Thanks much everyone. The video is now bright enough to be watchable.

[yetanotherlogin]

After editing the paths to the different binaries, I tried to convert the contents to an MP4 file using the x264 plug-in, but AviUtil isn't happy:

AviUtl: File export failed.

[C:\edited.mp4]
------------------------------------------------------------------------------------------------------------------------------
auo [info]: x264guiEx 2.24 / Windows 7 (x86) / AMD A8-3850 APU with Radeon(tm) HD Graphics
auo [info]: converting YUY2 -> nv12p, using SSE2
auo [info]: x264 options...
 --crf 23 --frames 528842 --input-res 720x576 --input-csp nv12 --fps 60000/1001 -o

"C:\edited.mp4" "-"
auo [error]: Failed to execute x264.
auo [info]: Total Encode Time:   : 0Hr  0Min  1.1Sec

Any idea why?

Thank you.

---
Edit: By default, AviUtil only has the 64-bit binaries:
https://forum.videohelp.com/threads/366724-Aviutl-Tips-Tricks-and-Support-thread/page13

ETA : 13h

[poisondeathray]

Did you edit the path AND rename the binary ?

If you can't get it working post a screenshot of the configuration screen

You can use a faster preset or encoding settings

AMD A8-3850 isn't ideal for encoding - it gets about 1/2 speed as i7 2600k
http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/44338-amd-a8-3850-apu-re...esktop-15.html

[yetanotherlogin]

Yes, it's running now that I downloaded the missing 32-bit binaries.

Been running 9h45; 9hrs left.

I know it's not a powerful computer, but the other tools I used so far (ffmpeg and Corel VideoStudio, primarily) were much faster so it wasn't a problem.

Should I kill the running task in AviUtil and use a faster preset?

[poisondeathray]

How long is the video? If we assume maybe 2 hours, 18-20 hours seems slow for SD video, even on that computer

What filters were you using, what encoding settings?

It's hard to say, unless you examine what the problem is. If you use some slow filter chain, and the bottleneck is the filter, using a faster preset won't really help. If the bottleneck is the CPU, then yes, using a faster preset will help

Corel VS and ffmpeg aren't valid comparisons, because presumably you weren't performing the same operations. If you test encoding only with each of those and aviutl , it would be more valid (apples to apples)

[yetanotherlogin]

The original video is this:

Input #0, mpeg, from 'full.mpeg':
  Duration: 02:26:46.52, start: 0.226767, bitrate: 8810 kb/s

    Stream #0:0[0x80]: Audio: ac3, 48000 Hz, 5.1(side), fltp, 448 kb/s

    Stream #0:1[0x1e0]: Video: mpeg2video (Main), yuv420p(tv), 720x576 [SAR 64:45 DAR 16:9], max. 9100 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc

I'm using all default settings, and the x264 encoder. Even with the "superfast" option, it's not any faster, so I guess AviUtil performs more CPU-intensive work than when transcoding using ffmpeg.

[poisondeathray]

The original video is this:

Input #0, mpeg, from 'full.mpeg':
  Duration: 02:26:46.52, start: 0.226767, bitrate: 8810 kb/s

    Stream #0:0[0x80]: Audio: ac3, 48000 Hz, 5.1(side), fltp, 448 kb/s

    Stream #0:1[0x1e0]: Video: mpeg2video (Main), yuv420p(tv), 720x576 [SAR 64:45 DAR 16:9], max. 9100 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc

I'm using all default settings, and the x264 encoder. Even with the "superfast" option, it's not any faster, so I guess AviUtil performs more CPU-intensive work than when transcoding using ffmpeg.

Were you testing encode only (no filters, just input/output) ? Filters eat up some CPU usage . For example if you use -vf filters such as a bunch of overlays , color correction etc.. it would be slower too in ffmpeg

[yetanotherlogin]

I just hit File > Export with plugin > Adv. x264 with default settings.

At the bottom, I guess it displays the settings it will pass on to the x264 encoder:

 --preset ultrafast --crf 23 --rc-lookahead 40 --aq-mode 1 --scenecut 40 --bframes 3 --b-adapt 1 --partitions p8x8,b8x8,i8x8,i4x4 --me hex --subme 7 --ref 3 --weightp 2 --trellis 1 --sar 64:45 --threads 2

[poisondeathray]

I'll look into it later in more detail, but if that was the line passed, the other commands will override the --preset . For example --preset ultrafast really means

                           --no-8x8dct --aq-mode 0 --b-adapt 0
                                    --bframes 0 --no-cabac --no-deblock
                                    --no-mbtree --me dia --no-mixed-refs
                                    --partitions none --rc-lookahead 0 --ref 1
                                    --scenecut 0 --subme 0 --trellis 0
                                    --no-weightb --weightp 0

But if you have other entries in the commandline they will override those preset settings. So that preset won't being doing much. You shouldn't use "ultrafast" in real life - the quality/compression ratio becomes worse than even xvid . "superfast" is the lowest you would go in real scenario

For your real project, your video was interlaced - so you need to either encode interlaced, or bob deinterlace and encode progressive

[yetanotherlogin]

Thanks, I'll check it out.

For a comparison, I wanted to also try the VirtualDub + AviSynth, but the following script 1) has no audio and 2) the video is streched vertically (16:9 to 4:3?):

#Unzipped DGDecode.dll from dgmpgdec158.zip into "c:\Program Files\AviSynth\plugins\"
#Ran DGIndex.exe, opened full.mpeg, and created full.d2v
Mpeg2Source("full.d2v") 
ColorYUV(cont_y=-20, gamma_y=200, off_y=-25)

What commands should I use instead?

Would those alternative commands work better, once they're changed to work with an MPEG file?

alpha=ImageSource("alpha.png", start=0, end=23, fps=23.976).ConvertToYV12(matrix="PC.601")
ImageSource("vlc.snapshot.bt.peniche.png", start=0, end=23, fps=23.976).ConvertToYV12()
darks=ColorYUV(gain_y=600, off_y=-30) # a lighted version of the video
brights=ColorYUV(cont_y=400, off_y=-170) # a darkened version of the video

Overlay(darks, brights, mask=alpha) # blend the two together

Mpeg2Source("full.d2v") 
ConvertToYV12()
HDRAGC(coef_gain=1.0, coef_sat=0.75)
HDRAGC(coef_gain=0.2, coef_sat=0.75)

Here's the input file:

c:\>ffprobe.exe -i full.mpeg
Input #0, mpeg, from 'full.mpeg':
  Duration: 02:26:46.52, start: 0.226767, bitrate: 8810 kb/s
    Stream #0:0[0x80]: Audio: ac3, 48000 Hz, 5.1(side), fltp, 448 kb/s
    Stream #0:1[0x1e0]: Video: mpeg2video (Main), yuv420p(tv), 720x576 [SAR 64:45 DAR 16:9], max. 9100 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc

Thank you.

[poisondeathray]

You still need to either encode interlaced (25i) or bob deinterlace and encode progressive (50p) . The example videos above by jagabo, and the HDRAGC example were bob deinterlaced to 50p

Your source uses "non-square pixels" . That's what 720x576 [SAR 64:45 DAR 16:9] means. It's "stretched" to 16:9 DAR (display aspect ratio) equivalent on playback. To do that you need to encode with --sar flags (--sar 64:45 in x264 or -vf setsar=sar=64/45 in ffmpeg). That's what the sample videos above used . Either that or resize to square pixel equivalents, which would be 1024x576 for "PAL" 16:9

To add audio in the avs script, you would load the audio and use AudioDub. When you use DGIndex, it demuxes the audio. If there is a delay value written into the name of the audio, you use DelayAudio() and whatever value in ms. There are more verbose instructions in the DGIndex manual.

e.g. it might look something like this (without filters)

mpeg2source("full.d2v")
vid=last

NicAc3Source("audio.ac3")
aud=last

Audiodub(vid,aud)

[yetanotherlogin]

Thanks much.

Alternatively, since ffmpeg also offers filters, I tried using this example from this page…

ffmpeg -i full.mpeg -c:v libx264 -vf mp=eq2=1:1.68:0.3:1.25:1:0.96:1 -c:a copy -f mp4 full.ffmpeg.mp4

… but ffmpeg complains with: ""[AVFilterGraph @ 049f8f60] No such filter: 'mp' Error opening filters!""

Could it be that it's not compiled with filters?

ffmpeg version N-73975-gda8b70b Copyright (c) 2000-2015 the FFmpeg developers
  built with gcc 4.9.2 (GCC)

  configuration: --enable-gpl --enable-version3 --disable-w32threads --enable-avisynth --enable-bzlib --enable-fontconfig --enable-frei0r --enable-gnutls --enab
le-iconv --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libdcadec --enable-libfreetype --enable-libgme --enable-libgsm --enable-l
ibilbc --enable-libmodplug --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-librtmp --en
able-libschroedinger --enable-libsoxr --enable-libspeex --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvo-aacenc --enable-libvo-amrwbenc
 --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs --enable-libxvid --enable-lzma --enable-decklink --enable-zlib

  libavutil      54. 28.100 / 54. 28.100
  libavcodec     56. 52.100 / 56. 52.100
  libavformat    56. 40.101 / 56. 40.101
  libavdevice    56.  4.100 / 56.  4.100
  libavfilter     5. 27.100 /  5. 27.100
  libswscale      3.  1.101 /  3.  1.101
  libswresample   1.  2.101 /  1.  2.101
  libpostproc    53.  3.100 / 53.  3.100

Thank you.

---

Edit : Apparently, ffmpeg removed the "mp" filter, and I should use the "-filter_complex" to split and modify the video into two parts, and join them.
https://ffmpeg.org/ffmpeg-filters.html

[yetanotherlogin]

You still need to either encode interlaced (25i) or bob deinterlace and encode progressive (50p) . The example videos above by jagabo, and the HDRAGC example were bob deinterlaced to 50p[code]

Thanks for the info.

Since it's working in AviUtil and I added all the slides, I'd rather use that option for now, and leave the AviSynth solution for other projects.

1. In AviUtil, how do I set the frame rate to 25fps? I set Filter+ > Change Framerate to "24 fps <- 30 fps", but when I select File > Export with plugin > Adv x264, it says "47.952fps".
2. OK How to make sure the picture won't be "pulled vertically"? In Filter+ > Deinterlace, I changed from "None" to "Auto" → Setting SAR to 64:45 does the job
3. Is there an easy solution to turn that 18hr job into 1-2hrs, besides selecting Filter+ > Compression > Video > Setting and moving the slider to the right, and, for audio compression, using MP3 in 128kbps 44.1kHz? ETA is still 9hrs, while Corel VideoStudio handles the same type of project in about 30mn

Thank you.

--
Edit:
Here's the output from the running console:

auo [info]: x264guiEx 2.24 / Windows 7 (x86) / AMD A8-3850 APU with Radeon(tm) HD Graphics
auo [info]: converting YUY2 -> nv12p, using SSE2
auo [info]: x264 version: 0.148.2597 e86f3a1tion: --bit-depth=8 --chroma-format=alluration: --bit-depth=8 --chroma-format=all
auo [info]: x264 options...
 --preset ultrafast --crf 23 --rc-lookahead 40 --aq-mode 1 --scenecut 40 --bframes 3 --b-adapt 1 --partitions
p8x8,b8x8,i8x8,i4x4 --me hex --subme 7 --ref 3 --weightp 2 --trellis 1 --sar 64:45 --threads 2 --frames 422544 --input-res
640x480 --input-csp nv12 --fps 48000/1001 -o "C:\test.mp4" "-"
raw [info]: 640x480p 64:45 @ 48000/1001 fps (cfr)
x264 [info]: using SAR=64/45
x264 [info]: using cpu capabilities: MMX2 SSE2Fast LZCNT
x264 [info]: profile Main, level 3.1

[jagabo]

18 hours for a 640x480 video encoded with x264 at the untrafast preset on a quad core CPU? The slowness must be in the filtering. On my computer (i5 2500K) a >2 hour movie from DVD with no filtering (aside from a resize) and those settings takes less than 10 minutes using the x264 cli encoder. So the encoding is a very small part of your overall conversion time.

Also, exporting at ~48 fps makes no sense. You're doing something wrong. If you inverse telecine from a 30 fps source you should have a 24 fps progressive video. If your source is interlaced and you bob deinterlace you should have a 60 fps video. If your source is interlaced and you encoded interlaced you should have a 30 fps video.