AVANTI - FFmpeg/Avisynth GUI (support thread).

[ricardouk]

It is filled with items in drop down menus that the program does not support.

For example...

And many of the templates split the audio file out.

Explain further...

Any attempt to set anything manualy and you get an error because most of the items in the drop down lists audio codecs, container etc are not supported by Avanti.

For example...

Were you looking for help? Give details of your problems... all you did was complain.

Avanti is not new, it's been around for a while (years) and its been tested by thousands, just because it doesnt work for you doesnt mean it doesn't work.

Explain which errors you're getting, which OS you're using, video source details, ffmpeg version etc.

[ricardouk]

after reading your other post here at videohelp i see what you mean but like Chris explained its for security reasons, you can overwrite those options, its all explained here:

https://forum.videohelp.com/threads/282223-AVANTI-FFmpeg-Avisynth-GUI-%28support-thread...=1#post1918982

[Chris K]

For those who like to play around with the latest FFmpeg changes/additions...

After the autobuilds closed and the FFmpeg developers dropped the SVN repository, it's a bit hard to keep up with the latest FFmpeg developments.

The last SVN revision was SVN-r26402 and the last available Win32 autobuild is SVN-r26400

http://ffmpeg.arrozcru.org/autobuilds/ffmpeg/mingw32/static

To still keep up-to-date with FFmpeg development (although not daily), you could try the Git builds provided at the following sites...

http://oss.netfarm.it/mplayer-win32.php
http://www.paehl.com/open_source/?Convert_Tools:FFMPEG
http://blog.k-tai-douga.com

Some notes about recent FFmpeg changes (nothing to do with SVN versus GIT)...

Pros:
Libavfilter now includes the "movie" filter which allows to overlay clips.
Input clip properties now shows the used profile for h264 sources.

Cons:
The "movie" filter works but doesn't support full Windows OS file paths yet.
The "wtv" decoder that works fine in SVN-r26400 now seems to be broken.

Avanti has always supported the "Sherpya" builds and more recently also the builds of Dirk Paehl. Both are slightly "off-generic" in the way they display the FFmpeg version info and Avanti needs some adaptions to use them.

I'll soon release Avanti 0.5.4 which fixes these and some other problems (h264 profile, -map_metadata syntax change). There will also be some adaptions to support libavfilter "labels" at the command line. These are needed to use e.g. the "movie" and "overlay" filters.

For the current and older Avanti versions, please stick to the SVN-r26400 revision.

Chris

[panaciq]

Hi there

I have one realy big problem with AVANTI: I am muxing one .jpg file and one .mp3 file into one .avi (Xvid) for uploading to YouTube, but YT gives constant error saying CANNOT CONVERT, so I look into .avi preferencies and there is one really strange thing - bitrate of the .avi is like 325297kbps?????????????????????? I could set anything in AVANTI as bitrate and it do not apply at all, it still remains this high! (( What do I do wrong? Also when I try to reconvert it with any other videoconverter it just remains too high, like multiplying the previous number (i.e.: now it is like 841524kbps)

EDIT: btw the same happens when I muxing .jpg+.mp3 with SUPER (video converter - but there I cannot set anything)

EDIT 2: suggestion - could you, please, list under codec the system ones too? Because when I now set Xvid as codec and hit START PROCESS Avanti complains "Unknown encoder 'xvid'" exiting with warnings (so no conversion/muxing is done at all), but there definitely is Xvid codec in my system, Avanti only acccepts Xvid (lib)...

EDIT 3: ok, now I noticed that it is probably not problem of AVANTI cos every converter I use sets the bitrate to such high number - I guess system for some reason sets bits not bytes - anyone know what is wrong here or where to set it as it should be? ((((

[panaciq]

...or maybe the problem is that when it is muxed (.jpg+.mp3 ---> .avi) it has only 1 FRAME??? Why it doesn't spread frames (that one frame like 200x) across whole avi to the length of the audio file that is muxed with the single jpg? I do not see any option for this in AVANTI (well actualy in any other converter/muxer I saw )

Therefore - I THINK - when I play it her ein my OS everything is OK but as soon as I try to uploa dit to YouTube they got problem with conversion to their online format (I guess they use .mp4 now)

[Chris K]

suggestion - could you, please, list under codec the system ones too? Because when I now set Xvid as codec and hit START PROCESS Avanti complains "Unknown encoder 'xvid'" exiting with warnings (so no conversion/muxing is done at all), but there definitely is Xvid codec in my system, Avanti only acccepts Xvid (lib)...

FFmpeg only uses its internal codecs. Both xvid and xvid (Lib) are on the list because very old FFmpeg versions expect xvid instead of libxvid. Perhaps I should remove it just like mp3 for which you now need to use mp3 (Lib). It's not recommended to use FFmpeg versions that old.

I know very little about Youtube but you could try to set the frame rate to 1 fps. Just overwrite the frame rate field and press enter (it then is stored in the list).

Then add -loop_input to the "User VIDEO options" command line. I don't know if Youtube accepts xvid/mp3 in avi but you can use above also with the x264/aac codecs. It will encode for the duration of the audio clip you have at "Source 2".

EDIT: You'll get a messagebox about the adio duration. You have to select "NO" there.

[Gavino]

...or maybe the problem is that when it is muxed (.jpg+.mp3 ---> .avi) it has only 1 FRAME??? Why it doesn't spread frames (that one frame like 200x) across whole avi to the length of the audio file that is muxed with the single jpg?

To repeat a single image frame, you need to use the -loop_input parameter to ffmpeg.
Unfortunately, this is a bit awkward in Avanti as it must come before the input file in the ffmpeg parameter list. This requires the use of -new in the "User VIDEO options" line in the Avanti GUI, which in turn means you have to add the other options explicitly - see the Avanti help for details.

A possible "User VIDEO options" line might be
-new -loop_input -vframes 4500 -i "@source1" -i "@source2"-dvs -das -y "@destin1"
This gives 4500 frames (3 mins @25fps) - change to suit your needs.

[Chris K]

To repeat a single image frame, you need to use the -loop_input parameter to ffmpeg.
Unfortunately, this is a bit awkward in Avanti as it must come before the input file in the ffmpeg parameter list. This requires the use of -new in the "User VIDEO options" line in the Avanti GUI, which in turn means you have to add the other options explicitly - see the Avanti help for details.

Hi Gavino,

You may have missed that but I recently added "command redirection" to the user command line. This is easier to use than the -new command. Just enter ...

-loop_input < [other options]

This will redirect the commands left from the < character to the start of the command line.

You can also use e.g. .... loop_input < [other options] > -newaudio

Everything right from the > character is redirected to after the "-y output" command.


Chris

[panaciq]

OK guys - I will try out...thanx a lot for your help!

[Gavino]

You may have missed that but I recently added "command redirection" to the user command line. This is easier to use than the -new command.

Thanks, Chris - that makes life a lot easier.

I missed the most recent updates - you work so fast I can't keep up.
Perhaps you could announce new releases on this thread to make users more aware.

@panaciq: as Chris says, you don't need the -vframes option as the duration will be taken from the audio input. Just use "-loop_input <", as Chris describes.

[panaciq]

...or maybe the problem is that when it is muxed (.jpg+.mp3 ---> .avi) it has only 1 FRAME??? Why it doesn't spread frames (that one frame like 200x) across whole avi to the length of the audio file that is muxed with the single jpg?

To repeat a single image frame, you need to use the -loop_input parameter to ffmpeg.
Unfortunately, this is a bit awkward in Avanti as it must come before the input file in the ffmpeg parameter list. This requires the use of -new in the "User VIDEO options" line in the Avanti GUI, which in turn means you have to add the other options explicitly - see the Avanti help for details.

A possible "User VIDEO options" line might be
-new -loop_input -vframes 4500 -i "@source1" -i "@source2"-dvs -das -y "@destin1"
This gives 4500 frames (3 mins @25fps) - change to suit your needs.

What is "-das" and "-dvs"? I did not find tgis two commands???

[panaciq]

well, so from now on AVANTI gives this error message:

-------------------------------------------------------------------------------------------------
11:29:17 - FFmpeg user VIDEO option(s) included.
11:29:17 - Free space on destination disk at start 4,176 Mb.

FFmpeg messages/warnings:

[mp3 @ 01cc0050] max_analyze_duration reached
[mp3 @ 01cc0050] Estimating duration from bitrate, this may be inaccurate

Metadata_Read:
... genre : Other

Unknown decoder 'libxvid'

11:29:17 - Process canceled by FFmpeg with messages/warnings.

-------------------------------------------------------------------------------------------------

What is wrong now?
It looks like it doesn't know it own internal codec? Why? Strange...

[Gavino]

What is "-das" and "-dvs"? I did not find tgis two commands???

They are not ffmpeg options but special flags for Avanti itself.
They tell Avanti to insert the "Destination Audio settings" and "Destination Video settings" from the GUI into the ffmpeg command line at that point, and are needed if you use the -new option. (See "The -new option" in the Avanti help file index.)

However, as Chris K explained, with Avanti 0.5.2 and above, you can use the much simpler command redirection feature instead of -new, and then you don't need -das and -dav. Just put
-loop_input <
at the start of your "User VIDEO options".

[panaciq]

================================================== ====================================

11:45:13 - FFmpeg user VIDEO option(s) included.
11:45:13 - Free space on destination disk at start 4,123 Mb.

FFmpeg messages/warnings:

[mp3 @ 01cc0050] max_analyze_duration reached
[mp3 @ 01cc0050] Estimating duration from bitrate, this may be inaccurate

Metadata_Read:
... genre : Other

[buffer @ 026f6b10] w:1024 h:768 pixfmt:yuvj444p
[scale @ 01d22330] w:1024 h:768 fmt:yuvj444p -> w:1024 h:768 fmt:yuv420p flags:0x4

Metadata_Write:
... ISFT : Lavf52.93.0
================================================== ====================================
WARNING! - FFmpeg showed no activity for more than 5 minutes (Cancel or temrinate process).


NO LUCK, GUYS! (((((((((((((

First it really started to do what it should and I was happy that it actualy do what it should (it creates .avi of the right approximate length it should have - cca. 15MB...and also it shows in AVANTI panel that it is doing what it should do - muxing the stuff) but at the end of the process it ends up with the above error...what is wrong now, please?

EDIT: so when I CANCEL the process and look for the outputed .avi and look into its properties there was even wrong bitrate (I set it in GUI like 500, but in propertis there was like 958 - why?)

BTW I guess process end up successfuly BUT anyway it put out that errror like it cannot temrinate it itself right...

[panaciq]

...and so after I CANCEL the process at the end it puts out this:
-----------------------------------------------------------------------------------------------------
12:18:00 - Process canceled by user (destination validity unknown).
12:18:02 - FourCC set to XVID, frame rate to 25.000 fps.

... Console processing <Ctrl><S> possibly shows more details.
-----------------------------------------------------------------------------------------------------

+ now I checked the outputed .avi: it has wrong length like 8 mins (it actualy should have 3:19 + bitrate this time I set in GUI to 1000 is shown as 1012) - I guess AVANTI ha sproblem detecting the length of the .mp3 (48kbps, 22kHz, mono)

EDIT: btw in GUI for AUDIO SETTINGS I set up "copy audio" as I do not want to be re-converted again, just leave it as it is

[Chris K]

It seems FFmpeg doesn't like the combination "mp3 + single jpeg + loop_input + acodec copy".

I tried the following...

1. loaded jpeg at "Source 1"
2. loaded mp3 at "Source 2"
3. loaded "DIVX / fast - 2-pass (using XVID codec)" template
4. Disabled 2-pass
5. Set bitrate to 500, framerate to 1 fps and audio to "Copy Audio"

I left off the -loop_input command (just leave command line as is).
When the audio duration message box appears, select "NO".

This way FFmpeg mux a clip with the duration of the mp3. Bitrate near 500 (464 kbps).
Because of the 1 fps frame rate, the file size is only slightly larger than the mp3 size.

If that doesn't work for youtube, then try the same procedure with 25 fps setting.

There are several threads about youtube upload here at the forums.
You can try the forum search option to find it (search for youtube in the title).

EDIT:

so when I CANCEL the process and look for the outputed .avi and look into its properties there was even wrong bitrate (I set it in GUI like 500, but in propertis there was like 958 - why?)

After I saw your destination frame size (1024x768). The higher bitrate means FFmpeg needs more than 500 kbps to follow its default quality criteria. If you want to force the 500 kbps then empty the command line and put -qmin 7 on it. You may have to increase the value slightly.

[panaciq]

It seems FFmpeg doesn't like the combination "mp3 + single jpeg + loop_input + acodec copy".

I tried the following...

1. loaded jpeg at "Source 1"
2. loaded mp3 at "Source 2"
3. loaded "DIVX / fast - 2-pass (using XVID codec)" template
4. Disabled 2-pass
5. Set bitrate to 500, framerate to 1 fps and audio to "Copy Audio"

I left off the -loop_input command (just leave command line as is).
When the audio duration message box appears, select "NO".

This way FFmpeg mux a clip with the duration of the mp3. Bitrate near 500 (464 kbps).
Because of the 1 fps frame rate, the file size is only slightly larger than the mp3 size.

If that doesn't work for youtube, then try the same procedure with 25 fps setting.

There are several threads about youtube upload here at the forums.
You can try the forum search option to find it (search for youtube in the title).

EDIT:

so when I CANCEL the process and look for the outputed .avi and look into its properties there was even wrong bitrate (I set it in GUI like 500, but in propertis there was like 958 - why?)

After I saw your destination frame size (1024x768). The higher bitrate means FFmpeg needs more than 500 kbps to follow its default quality criteria. If you want to force the 500 kbps then empty the command line and put -qmin 7 on it. You may have to increase the value slightly.

OK Chris, I will try that but beside this:

I SOLVED IT ANOTHER WAY AROUND (but your solution - if it will work - looks more like what I want, wiil try anyway): there need to be strictly defined "-vframes" (after "-loop_input < ") beacause AVANTI is unable to determine length from the .mp3

Now I am going to try your new solution...hope it'll work

EDIT: no, sorry but it doesn't work the way you suggested now ((( - bitrate is still shows bitrate like 12008kbps tho I did what you suggested (your numbered steps) with bitrate set in gui to 500kbps, except that: I have not problem muxing .jpg+.mp3 to .avi, the problem was much it to have it more than 1 FRAME IN VIDEO and this is exactly what happend with your suggestion - thanx anyway ...really up to now the most accurate setup is what I found out using your previous suggestion guys with my slight improvements and it works the best: set in GUI what I want to have (except bitrate as it do not work, I use "VBR qscale" instead), put "-loop_input < -vframes 200" cos as we now nknow AVANTi is unable to detemrine the length from the .mp3 file, so I check th elength myself, converted it to seconds only, put that value as "-vframes" value +1 (for not to be cut too sharp at the end of .mp3 but leave a bit space after last word) and finaly set in GUI Frame rate to 1.000 FPS - this way it is exactly as I want except the bitrate is somehow just guessed by VBR qscale (I am using value of 2, 3, or 4)

[panaciq]

These are actual settings that DO WORK for me:

[Chris K]

Yes, you're right! I checked it out with MediaInfo which says duration of video is 1 second which means only one frame with 1 fps setting.

But OK, it looks like you've set it to work, except for the high bitrate. I suggest to use higher values for the VBR qscale (5-10). This should lower the bitrate (and the quality of the converted jpeg).

EDIT:

Didn't see your screenshot yet when I posted but average bitrate of 696 doesn't look that bad.

[Gavino]

It seems FFmpeg doesn't like the combination "mp3 + single jpeg + loop_input + acodec copy".

Even re-encoding the audio (eg -acodec libmp3lame) gives the same problems.
It seems after the end of the audio stream is reached, ffmpeg just loops indefinitely processing the video.
I haven't tried with other forms of audio, but I suspect the same thing would happen.
Yesterday I said:

as Chris says, you don't need the -vframes option as the duration will be taken from the audio input. Just use "-loop_input <", as Chris describes.

It seems this isn't true and you do need -vframes.

[panaciq]

Didn't see your screenshot yet when I posted but average bitrate of 696 doesn't look that bad.

Exactly, Chris: I could set the highest posible value for VBR scale (=1) and the image quality is perfect and the whole size of video is under 10 MB with quite big 1024x768 (4:3) image resolution (which is the original size of the muxed image) - great!

Just that bitrate issue, aaaahhhh...but hey nevermind, hope you'll manage it somehow in next release of AVANTI...

[panaciq]

It seems FFmpeg doesn't like the combination "mp3 + single jpeg + loop_input + acodec copy".

Even re-encoding the audio (eg -acodec libmp3lame) gives the same problems.
It seems after the end of the audio stream is reached, ffmpeg just loops indefinitely processing the video.
I haven't tried with other forms of audio, but I suspect the same thing would happen.
Yesterday I said:

as Chris says, you don't need the -vframes option as the duration will be taken from the audio input. Just use "-loop_input <", as Chris describes.

It seems this isn't true and you do need -vframes.

@Gavino: exactly - I did try the same with the same wrong result (re-encoding the audio part, not just copy it)

[Chris K]

An alternative for the -vframes command would be to use the -t command, like...

-loop_input < -t 00:03:46 (duration of audio clip)

You then don't need to calculate second from the mp3 and it should also work with other frame rates.
You could copy the exact mp3 duration from the source properties.

I used a test clip that is reported by FFMpeg as 00:00:41.74 and copied that to the -t command

With this setting FFmpeg created a clip of 43 seconds where I expected 42 (one frame extra).

[panaciq]

An alternative for the -vframes command would be to use the -t command, like...

-loop_input < -t 00:03:46 (duration of audio clip)

You then don't need to calculate second from the mp3 and it should also work with other frame rates.
You could copy the exact mp3 duration from the source properties.

I used a test clip that is reported by FFMpeg as 00:00:41.74 and copied that to the -t command

With this setting FFmpeg created a clip of 43 seconds where I expected 42 (one frame extra).

@Chris: looks promising...will try that one

[panaciq]

@Chris: yes, confirmed - using "-t" parameter I achieved absolutely the same result as with "-vframes", so now it is OK (tho setting exact bitrate would be also great if it would work) - thanx a lot guys for helping me to solve this problem (see attached pic below for my final functional settings)!

[TestDemo]

Hello

I have 2 questions.

1. How do you use AVANTI to enocde to lagarith. I installed lagarith, updated the database then added it to the list of dropdowns in the gui just like in the help file, but when I tried to encode it says:

Unknown encoder 'lagarith'

2. I tried to encode to huffyuv(not the ffmpeg preset), but for come reason the output colourspace was 4:2:2, even though the input was 4:2:0. How does one encode to the input colourspace?

Would one say the output of huffyuv is yuy2 and the only way to do lossless through AVANTI is to use FFV1?

[Chris K]

1. How do you use AVANTI to enocde to lagarith. I installed lagarith, updated the database then added it to the list of dropdowns in the gui just like in the help file, but when I tried to encode it says:

Unknown encoder 'lagarith'

FFmpeg only can use its internal codecs. It's a common misunderstanding about FFmpeg that it would be able to use installed vfw or DirectShow codecs. At the Avanti help menu you can call up the "Codecs" and "Formats" list of the FFmpeg version you use.

 D..... = Decoding supported
 .E.... = Encoding supported
 ..V... = Video codec
 ..A... = Audio codec
 ..S... = Subtitle codec
 ...S.. = Supports draw_horiz_band
 ....D. = Supports direct rendering method 1
 .....T = Supports weird frame truncation
 ------
 D V D  lagarith        Lagarith lossless
 DEVSD  huffyuv         Huffyuv / HuffYUV

Here you can see that FFmpeg only supports decoding of lagarith.

2. I tried to encode to huffyuv(not the ffmpeg preset), but for come reason the output colourspace was 4:2:2, even though the input was 4:2:0. How does one encode to the input colourspace?

Would one say the output of huffyuv is yuy2 and the only way to do lossless through AVANTI is to use FFV1?

I only know the basics of colorspace. I have the approach that if a step in the process can only output (or work, like Avisynth plugins) in a certain colorspace, I'll have to live with that. As a GUI developer I'm more concerned about interfacing with FFmpeg than the details of every individual codec. Sorry for that.

[Gavino]

The recent question about looping a single jpg has reminded me of a related issue I've been meaning to ask about for a while.
Are there any plans to include support in Avanti for image sequences such as img%03d.jpg as input?
ffmpeg supports this, but Avanti rejects it as it does not recognise the file name as valid.

It's not a high priority (you can always use an Avisynth script with ImageSource instead), but it would be a 'nice-to-have' feature.

[panaciq]

The recent question about looping a single jpg has reminded me of a related issue I've been meaning to ask about for a while.
Are there any plans to include support in Avanti for image sequences such as img%03d.jpg as input?
ffmpeg supports this, but Avanti rejects it as it does not recognise the file name as valid.

It's not a high priority (you can always use an Avisynth script with ImageSource instead), but it would be a 'nice-to-have' feature.

yea, I am for that too - it'd be really nice option to have right in AVANTI GUI

[Chris K]

Are there any plans to include support in Avanti for image sequences such as img%03d.jpg as input?
ffmpeg supports this, but Avanti rejects it as it does not recognise the file name as valid.

I'll take a look into it. One of the problems is that Avanti collects the source properties when you load a file so something abstract as "img%03d.jpg" will throw an error. Perhaps just load the first image file like img000.jpg or img001.jpg and flag in some way it concerns a image sequence.

I can't use the % character as a key because Windows allows % in file names. E.g.; "test%video%clip.avi" is a valid file name and even "img%03d.jpg" would be valid if you really renamed a jpeg to this.

But we'll see! I guess it can be done, but with some restrictions on progress display because the duration of the source is hard to predict, unless you know the number of images to expect.