trouble adding a different script to a trimmed section in my script

[unclescoob]

My script consists of various trimmed sections, as I am applying different filters to different sections in my video.


So for example, something like this works fine:

vid2= trim (2485, 3637). TTempSmooth(maxr=7,lthresh=20,cthresh=5, strength=5, interlaced=false)

But there is a specific section that I am applying flatmasks to, and I'm getting a syntax error. The script below is the one I'm using in that particular section. Please advise.

Thanks

MPEG2Source(video.d2v)
AssumeTFF()
TFM()
TDecimate(mode=1)
Crop(6, 0, -6, -0)
##I am not including vid 1, vid 2, etc. in this example, just the vid I'm having trouble with. ###

vid4= trim (5435,5614).  filtered=trim. TTempSmooth(maxr=7,lthresh=20,cthresh=5, strength=5, interlaced=false). flatmask(filtered, str=100, r=10, luma=true, lth=32, upper=true, show=false)

[manono]

MPEG2Source(video.d2v) 
AssumeTFF() 
TFM() 
TDecimate(mode=1) 
Crop(6, 0, -6, -0) ##I am not including vid 1, vid 2, etc. in this example, just the vid I'm having trouble with. ###  
vid4= trim (5435,5614).  filtered=trim. TTempSmooth(maxr=7,lthresh=20,cthresh=5, strength=5, interlaced=false). flatmask(filtered, str=100, r=10, luma=true, lth=32, upper=true, show=false)

Huh? Do you mean to say filtered=vid4? Trim, as used in the bold-faced section, is undefined. I don't know if anything's wrong with the rest of it as I don't understand some of it.

[unclescoob]

I tried to do it with filtered=vid4, and I get a syntax error.

My original script for that clip only goes as follows:

MPEG2Source("video.d2v")
TFM() 
TDecimate(mode=1) 
filtered=last
TTempSmooth(maxr=7,lthresh=20,cthresh=5, strength=5, interlaced=false)
flatmask(filtered, str=100, r=10, luma=true, lth=32, upper=true, show=false)
McTemporalDenoise(settings="high")
Tweak(hue=0.0, sat=0.80, bright=0, cont=1.0, coring=true, sse=false)
Import("C:\Program Files\AviSynth 2.5\plugins\fastlinedarkenmod.avs")
Fastlinedarkenmod(strength=60)

How can I incorporate this same script into a trimmed section of my video? Note, my entire video by default is filtered by MCTemporalDenoise. But certain sections (like the one discussed) need additional filters. This one for instance needs flatmasks, because I only wanted TTempSmooth to filter a specific section in that clip, but not the entire thing.

[manono]

How can I incorporate this same script into a trimmed section of my video?

In my opinion, the best way to filter parts of a video is with the RemapFrames.dll by stickboy, of which ReplaceFramesSimple is a part. You don't need to use Trim at all.

You get it here:

http://avisynth.org/stickboy/

And you use it like this:

BaseClip=Last
SourceClip=BaseClip
SourceClip=SourceClip.filter1.filter2.filter3...
ReplaceFramesSimple(BaseClip,SourceClip,Mappings="[FirstFrame1 LastFrame1] [FirstFrame2 LastFrame2] [FirstFrame3 LastFrame3]")

You can use it multiple times in a video to filter different parts differently.

That doesn't answer your original question, of course, but one way to figure where it's going wrong when the error message doesn't point to a specific line is to put a 'Return Last' after one line, open it in VDub to check, and then move the 'Return Last' down line-by-line until you come to the offending line.

[unclescoob]

"Return last" results in the entire mask to apply to the rest of the clip, which is not what I want.

Update: the mask section of my script works, but it applies the mask to the rest of my clip from that point on. That shouldn't happen.

Here is a sample of my script.

##starting the script with the basics--the video with simple IVTC script##
MPEG2Source(video.d2v)
AssumeTFF()
TFM()
TDecimate(mode=1)
Crop(6, 0, -6, -0)

#from frame 0-300- apply no filters#
vid1=Trim(0,300)


#from frame 301 to frame 400, apply grayscale#
vid2=Trim(301,400). Grayscale()

#from 401 - 500, back to color (no other filters)#
vid3=trim(401,500)

#from 501-900 apply the mask#
vid4=trim(501,900)
Raw=last
Invert()
flatmask(raw, str=100, r=10, luma=true, lth=32, upper=true, show=false)

##from 901-1000 apply no filters##  ###<-------although I trimmed vid5 separately, flatmasks from vid4 is still being applied to this.  Why???###
vid5=trim(901,1000)

vid1 ++vid2 ++vid3 ++vid4 ++vid5

[manono]

"Return last" results in the entire mask to apply to the rest of the clip, which is not what I want.

The suggestion was to use 'Return Last' as a test to see where the script is going wrong - to figure out which line is responsible for the error, not to encode with it on. But now that I see the script, I realize the suggestion might not work.

What happens if you move:
vid5=trim(901,1000)

above the part beginning with:

vid4=trim(501,900)

If that doesn't help at all, then smarter people than I will have to help out.

[unclescoob]

It worked. How did reversing the order work?

[poisondeathray]

Because "last" is implied when you don't mention otherwise. It refers to the section previously

Since vid5=trim(901,1000) was preceded by the already filtered vid4 section, that's the new "last" . It no longer applies to the main unfiltered video



If it's more clear, you can label the sections more clearly, which is the main video etc...

[unclescoob]

Ah, got it. Thanks fellas.

[poisondeathray]

vid5=trim(901,1000) really means vid5=last.trim(901,1000)

I break it out and label everything

So instead of omitting "last" , I use real names and labels. I might use "mainvideo" , or even other names like "a" or "b" , whatever is appropriate because I get confused easily and when you have huge scripts with many sections it gets hectic quickly




This section actually doesn't do what you think it does, it's applied to the whole video

#from 501-900 apply the mask
vid4=trim(501,900)
Raw=last
Invert()
flatmask(raw, str=100, r=10, luma=true, lth=32, upper=true, show=false)



But raw=last means now you call the IVTCed cropped section "Raw" now (instead of "last). It's not applied to the "vid4" label. So the entire video is inverted and flatmasked because "last" was the IVTCed and cropped main video . And when you use vid5 in the next section, last is implied, so it calls that entire flatmasked video


This section could be re-written

Raw=last
Invert()
flatmask(raw, str=100, r=10, luma=true, lth=32, upper=true, show=false)
Trim(501,900)
vid4=last

So now "vid4" refers to that flatmasked 501-900 section

[unclescoob]

This section actually doesn't do what you think it does, it's applied to the whole video

#from 501-900 apply the mask
vid4=trim(501,900)
Raw=last
Invert()
flatmask(raw, str=100, r=10, luma=true, lth=32, upper=true, show=false)

But raw=last means now you call the IVTCed cropped section "Raw" now (instead of "last). It's not applied to the "vid4" label. So the entire video is inverted and flatmasked because "last" was the IVTCed and cropped main video . And when you use vid5 in the next section, last is implied, so it calls that entire flatmasked video

In other words, "Last" means the last video segment rendered to the script. In this case, MPEG2Source(). Therefore, the mask will be applied to the entire video, ignoring trimmed sections. Am I on track here?

This section could be re-written

Raw=last
Invert()
flatmask(raw, str=100, r=10, luma=true, lth=32, upper=true, show=false)
Trim(501,900)
vid4=last

So now "vid4" refers to that flatmasked 501-900 section

This works well, except now, if I add additional trimmed sections following the masked trimmed section, it won't show up. My clip will end at frame 900. Manono's suggestion of reversing the order: adding video 5,6,7, etc after video 3 and leaving video 4 on the last line (hence "last" in the function) worked.

[poisondeathray]

This section actually doesn't do what you think it does, it's applied to the whole video

#from 501-900 apply the mask
vid4=trim(501,900)
Raw=last
Invert()
flatmask(raw, str=100, r=10, luma=true, lth=32, upper=true, show=false)

But raw=last means now you call the IVTCed cropped section "Raw" now (instead of "last). It's not applied to the "vid4" label. So the entire video is inverted and flatmasked because "last" was the IVTCed and cropped main video . And when you use vid5 in the next section, last is implied, so it calls that entire flatmasked video

In other words, "Last" means the last video segment rendered to the script. In this case, MPEG2Source(). Therefore, the mask will be applied to the entire video, ignoring trimmed sections. Am I on track here?

Yes, the MPEG2Source() with the IVTC and crop . Not the original unfiltered video

This section could be re-written

Raw=last
Invert()
flatmask(raw, str=100, r=10, luma=true, lth=32, upper=true, show=false)
Trim(501,900)
vid4=last

So now "vid4" refers to that flatmasked 501-900 section

This works well, except now, if I add additional trimmed sections following the masked trimmed section, it won't show up. My clip will end at frame 900. Manono's suggestion of reversing the order: adding video 5,6,7, etc after video 3 and leaving video 4 on the last line (hence "last" in the function) worked.

I was trying to explain (poorly) that you should use descriptors for each section . If you add additional trimmed sections, don't use implicit "last" ; use whatever name you assigned to the MPEG2Souce , IVTC, crop section . Otherwise, you have to place all your unfiltered sections before all your masked sections in script order or you will run into the same problems as before (because "last" will no longer refer to the MPEG2Source) . When you begin to apply different filters to different sections (more complex scripts), it will be difficult to keep track

If you're only doing this type of masking - using replaceframes, replaceframesimple as suggested above, or clipclop would easier and more elegant solution than using trims

[unclescoob]

Of course, replaceframes is just a matter of choice for the sake of simplifying the look of your script, right? It doesn't affect the framerate or anything in terms of quality?

[poisondeathray]

Of course, replaceframes is just a matter of choice for the sake of simplifying the look of your script, right? It doesn't affect the framerate or anything in terms of quality?

Yes

But it's faster for writing (fewer keys to type) , and it's easier to organize complex scripts - you can mix & match different sections more easily in much less time (to type) . Think of it as mixing & matching different versions of the video

Note - there are multiple different ways of organizing scripts. You'll eventually find the "style" that you prefer

The danger in not using labels or organizing is "last" or "implicit last" might not refer to what you think it refers to and you get double filtered and unexpected results

[unclescoob]

Quick question off topic: Is transcoding mpeg2 to h.264 with I and P frames only a better option for quality, IF your goal is to lose as little quality as possible in the transcode process? (I know this should be in a different thread, just hoping it's a simple yes, no, usually yes, usually no)

[poisondeathray]

Quick question off topic: Is transcoding mpeg2 to h.264 with I and P frames only a better option for quality, IF your goal is to lose as little quality as possible in the transcode process? (I know this should be in a different thread, just hoping it's a simple yes, no, usually yes, usually no)

It's not a simple answer

If you use very low quantizers (very high bitrates) relative to content complexity , then using all I frames will give higher quality . But I frames cost a lot of bitrate (they are encoded separately , and do not derive any information from other frames)

p and b-frames give better temporal compression, but are slighly lower in quality (you normally don't see the difference in normal viewing)

In "normal" bitrate ranges, using a mix of p and b frames will generally give higher quality overall

Some h.264 encoders can encode lossless e.g. x264. Even lossless with p and b frames (long gop lossless), but lossless h.264 isn't supported by most hardware (only software)

[unclescoob]

Thanks. And thanks for the trim info.

[unclescoob]

PDR, could you give me a script sample of what you mean? I think the visual will help me alot in conjunction with your explanation. Perhaps you can just re-arrange my script to reflect your explanation?

Thanks

[poisondeathray]

PDR, could you give me a script sample of what you mean? I think the visual will help me alot in conjunction with your explanation. Perhaps you can just re-arrange my script to reflect your explanation?

I'm assuming you mean just broken out? Not replaceframes/replaceframessimple? - because there is an example above of that and many examples in the documentation


MPEG2Source()
TFM
TDecimate
Crop
Main = last

#Now "main" refers to the entire video IVTCed and cropped. So instead of using "last" or "implied last". You call it by using main with your trimmed sections


Main
filter1()
filter2()
Trim(0,100)
a=last

#Now "a" refers to "Main" with filter1 and filter2 applied . It's been cut to 0-100 .


Main
filter3()
filter4()
Trim(101,200)
b=last

#Now "b" refers to "Main" with filter3 and filter 4 applied. It's been cut to frames 101-200

a++b


Notice they are always named variables. Everything is spelled out. (You explicitly call "Main" to start so you know exactly what to expect). You don't have to use "a" or "b" . You can use other variable name like "dog" or "cat". Or maybe use something more descriptive for that section like "darksection_filtered". Notice "last" is never implied, or used the way you used it earlier (that's where you can make mistakes if you don't keep track of what "last" really means) . When you use the notation like vid8=trim(x,y) , you don't know video you're referring to, so "last" is implied. But you changed what "last" meant when doing the mask section


This is a tedious way to do , but you're less prone to mistakes . ReplaceFrames or similar functions is the better way to do it in avisynth IMO. They allow you to repeat filters in alternate sections much easier instead of typing everything out multiple times (if you're applying the same set of filters over different sections, and other sets of filters over other sections) . It can be many times faster to do it that way (typing wise, not processing speed)

[unclescoob]

Beautiful. I'll give this a shot this afternoon.

Thanks!!

[unclescoob]

Worked like a charm. And yeah, I prefer to use trim for now. It may require more typing, but I'm used to it. However, your sample makes my script look alot cleaner, more organized and easier to work with. Thanks again, folks.