Aviutl Tips, Tricks and Support thread

[PhotoCat]

I tried to set the max system video size to 5616x5616 and it almost worked to let me zoom the
image to 1:1 but aviutl ran out of memory quickly and crashed. I guess I really need to pre-crop the image.

[PhotoCat]

X/Y rectangular movement - any shortcut to swap the start and end positions?

Same question for zoom start & end. Tks!

[racer-x]

You need to understand that you are working with a 32-bit program, meaning limited memory management. I have my settings @ 4096 x 2304 to deal with 4k footage. That means my images can be no more than 2304 in any dimension.

"X/Y rectangular movement - any shortcut to swap the start and end positions?"

Try positive / negative values.

[MaverickTse]

High resolution jpg file in slideshow. How to zoom to 1:1 in high quality?

I have imported a jpg file of size 5616x3744 using Maverick's auto shrink plugin. However,
I have discovered that when I zoomed the image in aviutl to 100%, the quality was significantly
worse than my image viewer (faststone) in 100% view (ie less detail, less sharp). Is there anyway to preserve
the original image quality in aviutl when zoomed in?

I am guessing the auto shrink plugin will just shrink down the image file into a smaller sized image that
aviutl can handle, and therefore zooming with aviutl of the same image will not bring back the original
image quality. Not sure if this is the reason...

Thinking about a workaround... Should I just crop the original image in the face area rather than shrink it... ?

Anything I am missing? Pls advise! Thanks!

Yeah, that's right, I just shrink whatever oversized image to what AviUtl can handle at current settings.
The size that you just loaded the image is ALREADY the MAXIMUM AviUtl can handle properly. If you magnify it, even by 1 pixel, the usual problems occur.

Ask your self again do you really need such resolution. Are you making an A0 poster instead of making a video?

5616x3744, when expanded in memory, leads to a 80MB RGBA data block. Considering the intermediate shrunken image, you can take it as 100MB.

If you need an oversized background while only need to show part of it at a time, try using the SVG Custom object.
First embed you image in SVG using Adobe Illustrator or InkScape (If using Inkscape, do not apply any deformation);
Then add "Adv@svg-render-EN", point it to your SVG file.

[MaverickTse]

X/Y rectangular movement - any shortcut to swap the start and end positions?

Same question for zoom start & end. Tks!

Click on the number then type in directly

[poisondeathray]

are there any plans for an x64 version in the future ?

[PhotoCat]

High resolution jpg file in slideshow. How to zoom to 1:1 in high quality?

Tks Maverick for your reply!

I have imported a jpg file of size 5616x3744 using Maverick's auto shrink plugin. However,
I have discovered that when I zoomed the image in aviutl to 100%, the quality was significantly
worse than my image viewer (faststone) in 100% view (ie less detail, less sharp). Is there anyway to preserve
the original image quality in aviutl when zoomed in?

I am guessing the auto shrink plugin will just shrink down the image file into a smaller sized image that
aviutl can handle, and therefore zooming with aviutl of the same image will not bring back the original
image quality. Not sure if this is the reason...

Thinking about a workaround... Should I just crop the original image in the face area rather than shrink it... ?

Anything I am missing? Pls advise! Thanks!

Ask your self again do you really need such resolution. Are you making an A0 poster instead of making a video?

5616x3744, when expanded in memory, leads to a 80MB RGBA data block. Considering the intermediate shrunken image, you can take it as 100MB.

If you need an oversized background while only need to show part of it at a time, try using the SVG Custom object.
First embed you image in SVG using Adobe Illustrator or InkScape (If using Inkscape, do not apply any deformation);
Then add "Adv@svg-render-EN", point it to your SVG file.

No, I don't need an A0 poster. All I need is 1080P video. However, it would be convenient to be able to zoom to
1:1 of the 5616x3744 image and use only a small crop of it. (e.g. 1920x1080 crop) This way, one can zoom to the
face of a full length portrait with ease in aviutl. Right now I have to make a crop on the face and show that as a video/slideshow
segment. Otherwise, the face details would be gone.

Anyway to crop the full sized image as one moves the image around in aviutl?

SVG won't work as it is a photograph. Perhaps something like a "smart object" in photoshop would work??

No, I don't need the full res image 5616x3744 at all times. I just need a crop of 1920x1080 of the full-sized image whenever needed.

Thanks!

[PhotoCat]

X/Y rectangular movement - any shortcut to swap the start and end positions?

Same question for zoom start & end. Tks!

Click on the number then type in directly

Done that. Hv to type in 3 values X, Y & Zoom most of the time.

It would be nice to be able to hit a button and copy all the start data column to the finish column. (X, Y, Z & Zoom etc)
This quick copy would allow us to enter mid-point data quickly.

Also a nice short cut key to swap the start column with the finish column. This would allow us
to quickly change a zoom-in effect to a zoom-out effect in a slideshow.

Thanks Maverick!

[MaverickTse]

are there any plans for an x64 version in the future ?

I do not own the code of AviUtl.exe and Exedit.auf, and I do not get any news on x64 version.
but I would try to press the author to release his code if he abandon development completely.

[MaverickTse]

High resolution jpg file in slideshow. How to zoom to 1:1 in high quality?

Tks Maverick for your reply!

I have imported a jpg file of size 5616x3744 using Maverick's auto shrink plugin. However,
I have discovered that when I zoomed the image in aviutl to 100%, the quality was significantly
worse than my image viewer (faststone) in 100% view (ie less detail, less sharp). Is there anyway to preserve
the original image quality in aviutl when zoomed in?

I am guessing the auto shrink plugin will just shrink down the image file into a smaller sized image that
aviutl can handle, and therefore zooming with aviutl of the same image will not bring back the original
image quality. Not sure if this is the reason...

Thinking about a workaround... Should I just crop the original image in the face area rather than shrink it... ?

Anything I am missing? Pls advise! Thanks!

Ask your self again do you really need such resolution. Are you making an A0 poster instead of making a video?

5616x3744, when expanded in memory, leads to a 80MB RGBA data block. Considering the intermediate shrunken image, you can take it as 100MB.

If you need an oversized background while only need to show part of it at a time, try using the SVG Custom object.
First embed you image in SVG using Adobe Illustrator or InkScape (If using Inkscape, do not apply any deformation);
Then add "Adv@svg-render-EN", point it to your SVG file.

No, I don't need an A0 poster. All I need is 1080P video. However, it would be convenient to be able to zoom to
1:1 of the 5616x3744 image and use only a small crop of it. (e.g. 1920x1080 crop) This way, one can zoom to the
face of a full length portrait with ease in aviutl. Right now I have to make a crop on the face and show that as a video/slideshow
segment. Otherwise, the face details would be gone.

Anyway to crop the full sized image as one moves the image around in aviutl?

SVG won't work as it is a photograph. Perhaps something like a "smart object" in photoshop would work??

No, I don't need the full res image 5616x3744 at all times. I just need a crop of 1920x1080 of the full-sized image whenever needed.

Thanks!

Nope. Trust me. You CAN embed bitmap in a SVG file. I have tried with smaller bitmaps (1280x720), uncertain about the larger one.
You can also save the SVG as SVGZ (zipped) to save some space.

[PhotoCat]

You need to understand that you are working with a 32-bit program, meaning limited memory management. I have my settings @ 4096 x 2304 to deal with 4k footage. That means my images can be no more than 2304 in any dimension.

"X/Y rectangular movement - any shortcut to swap the start and end positions?"

Try positive / negative values.

Tks racer!

[PhotoCat]

Created a new project in 1920x1080 but the project became 1280x720!

On top of the window, it says, 1920x1080_30fps_44100Hz.exedit (1280,720) [1/1]*...

I think it worked b4. Anything I am doing wrong? Tks!

Anyone knows the answer for this one? Perhaps I need to re-install aviutl...? Tks!

[racer-x]

Looks like you have the project set to resize to 1280 x 720.

Click: Filter+ > Change Size. Make sure "None" is selected.

[PhotoCat]

U r right on the money Racer! Thank you so much

[PhotoCat]

Learning Camera Control

Found Maverick's tutorial here:

https://www.youtube.com/watch?v=WOwQiY8Odn8

and have a few questions:

1) Why do we hv to set "target ly" to the CC layer number? What does it do?

I found X, Y & Z behave quite differently if u set "target ly" correctly and incorrectly.

What is the theory behind "target ly"??

2) X, Y, Z & Center X, Y, Z seem to be doing the same thing??

Anyway happy Black Friday!

[MaverickTse]

target ly=0 -> fly-by origin mode (always face the origin 0,0,0)
target ly= CameraControl -> Pan and Zoom mode
target ly= object -> track on object/fly-by object mode (always face towards object)

[MaverickTse]

CenterX/Y/Z = Anchor point (try to rotate the object after you change this)

[MaverickTse]

What a BLACK Friday! I gave a LECTURE to my publisher for doing a lousy job (and they don't aware of their problem)... I can't recall when the last time I gave people a lecture...
Literally I turned BLACK this day

[PhotoCat]

CenterX/Y/Z = Anchor point (try to rotate the object after you change this)

got it! thanks!

[PhotoCat]

target ly=0 -> fly-by origin mode (always face the origin 0,0,0)
target ly= CameraControl -> Pan and Zoom mode
target ly= object -> track on object/fly-by object mode (always face towards object)

Kind of got it... thanks! Chill out man... it would be a great book when it is all done...

[Marchand]

Test: AviUtl-ENG1.8Extra[Win7] - L-SMASH rev1316

x265 1.4+144-d32249002258:[Windows][GCC 4.6.3][64 bit] 16bpp
x265 [info]: Main 10 profile, Level-4 (Main tier)

[C:\Users\Cracker_Marchand\Videos\teste avutli.mp4]
------------------------------------------------------------------------------------------------------------------------------
auo [info]: x265guiEx 3.49 / Windows 8.1 (x64) / Intel Core2 Quad Q8400 @ 2.66GHz (4C/4T)
auo [info]: converting YC48 -> i420p(16bit), using SSE2 SSSE3 SSE4.1
auo [info]: x265 options...
--input-depth 16 --input-csp i420 --preset faster --bitrate 1500 --frames 179 --input-res 1920x1080 --fps 1990/83 -o
"C:\Users\Cracker_Marchand\Videos\teste avutli.265" "-"
yuv [info]: 1920x1080 fps 1990/83 i420p16 unknown frame count
x265 [info]: HEVC encoder version 1.4+144-d32249002258
x265 [info]: build info [Windows][GCC 4.6.3][64 bit] 16bpp
x265 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.1 Cache64
x265 [info]: Main 10 profile, Level-4 (Main tier)
x265 [info]: WPP streams / frame threads / pool : 17 / 2 / 4
x265 [info]: Internal bit depth : 10
x265 [info]: CTU size / RQT depth inter / intra : 64 / 1 / 1
x265 [info]: ME / range / subpel / merge : hex / 57 / 2 / 2
x265 [info]: Keyframe min / max / scenecut : 23 / 250 / 40
x265 [info]: Lookahead / bframes / badapt : 15 / 4 / 0
x265 [info]: b-pyramid / weightp / weightb / refs: 1 / 1 / 0 / 1
x265 [info]: Rate Control / AQ-Strength / CUTree : ABR-1500 kbps / 1.0 / 0
x265 [info]: tools: rd=2 early-skip deblock sao signhide fast-intra tmvp
x265 [info]: frame I: 3, Avg QP:23.70 kb/s: 1931.88
x265 [info]: frame P: 42, Avg QP:28.98 kb/s: 2757.57
x265 [info]: frame B: 134, Avg QP:31.96 kb/s: 825.46
x265 [info]: global : 179, Avg QP:31.13 kb/s: 1297.35
x265 [info]: Weighted P-Frames: Y:23.8% UV:21.4%
x265 [info]: consecutive B-frames: 22.2% 4.4% 0.0% 0.0% 73.3%

encoded 179 frames in 47.36s (3.78 fps), 1297.35 kb/s
auo [info]: CPUŽg—p—¦: Aviutl: 9.91% / x265: 82.45%
auo [info]: x265ƒGƒ“ƒR�[ƒhŽžŠÔ : 0ŽžŠÔ 0•ª47.5•b
auo [info]: NeroAacEnc ‚ʼn¹�ºƒGƒ“ƒR�[ƒh‚ð�s‚¢‚Ü‚·�B AAC-LC ƒrƒbƒgƒŒ�[ƒgŽw’è, 128kbps
auo [info]: L-SMASH muxer (r1316) ‚Åmux‚ð�s‚¢‚Ü‚·�B‰f‘œ: on, ‰¹�ºff, tcff, chapff, Šg’£ƒ‚�[ƒh:‚È‚µ
auo [info]: L-SMASH remuxer (r1316) ‚Åmux‚ð�s‚¢‚Ü‚·�B‰f‘œ: on, ‰¹�º: on, tcff, chapff, Šg’£ƒ‚�[ƒh:‚È‚µ
auo [info]: ‘�ƒGƒ“ƒR�[ƒhŽžŠÔ : 0ŽžŠÔ 0•ª48.9•b

Geral
Nome completo : C:\Users\Cracker_Marchand\Videos\teste avutli.mp4
Formato : MPEG-4
ID do Codec : M4V
Tamanho do arquivo : 1.27 MiB
Duração : 7s 530ms
Taxa de Bits Total, Modo : Variável
Taxa de Bits Total : 1 414 Kbps
Data da codificação : UTC 2014-11-29 00:14:55
Data rotulada : UTC 2014-11-29 00:14:55
Programa usado : L-SMASH

Vídeo
ID : 1
Formato : HEVC
Formato/Informações : High Efficiency Video Coding
Perfil do Formato : Main 10@L4.0
ID do Codec : hvc1
ID do Codec/Informações : High Efficiency Video Coding
Duração : 7s 465ms
Taxa de bits : 1 299 Kbps
Largura : 1 920 pixels
Altura : 1 080 pixels
Proporção da imagem : 16:9
Modo de taxa de quadros : TConstante
Taxa de quadros : 23.976 fps
ColorSpace : YUV
ChromaSubsampling : 4:2:0
BitDepth/String : 10 bits
Bits/(Pixels*Quadros) : 0.026
Tamanho da Faixa : 1.16 MiB (91%)
Biblioteca usada : x265 1.4+144-d32249002258:[Windows][GCC 4.6.3][64 bit] 16bpp
Configurações de codificação : wpp / ctu=64 / tu-intra-depth=1 / tu-inter-depth=1 / me=1 / subme=2 / merange=57 / no-rect / no-amp / max-merge=2 / temporal-mvp / early-skip / no-fast-cbf / rdpenalty=0 / no-tskip / no-tskip-fast / strong-intra-smoothing / no-lossless / no-cu-lossless / no-constrained-intra / fast-intra / open-gop / interlace=0 / keyint=250 / min-keyint=23 / scenecut=40 / rc-lookahead=15 / bframes=4 / bframe-bias=0 / b-adapt=0 / ref=1 / weightp / no-weightb / aq-mode=2 / aq-strength=1.00 / cbqpoffs=0 / crqpoffs=0 / rd=2 / psy-rd=0.00 / psy-rdoq=0.00 / signhide / lft / sao / no-sao-non-deblock / b-pyramid / no-cutree / rc=abr / bitrate=1500 / ratetol=1.0 / qcomp=0.60 / qpmin=0 / qpmax=51 / qpstep=4 / ipratio=1.40 / pbratio=1.30
Data da codificação : UTC 2014-11-29 00:14:55
Data rotulada : UTC 2014-11-29 00:14:55

Áudio
ID : 2
Formato : AAC
Formato/Informações : Advanced Audio Codec
Perfil do Formato : LC
ID do Codec : 40
Duração : 7s 530ms
Modo da taxa de bits : Variável
Taxa de bits : 120 Kbps
Nº de canais : 2 canais
Posições dos canais : Front: L R
Taxa de amostragem : 48.0 KHz
Tamanho da Faixa : 110 KiB (8%)
Biblioteca usada : Nero AAC codec 1.5.4.0
Configurações de codificação : -br 128000 -lc
Data da codificação : UTC 2014-11-29 00:14:55
Data rotulada : UTC 2014-11-29 00:14:55

[MaverickTse]

Test: AviUtl-ENG1.8Extra[Win7] - L-SMASH rev1316

x265 1.4+144-d32249002258:[Windows][GCC 4.6.3][64 bit] 16bpp
x265 [info]: Main 10 profile, Level-4 (Main tier)

[/QUOTE]

I am aware of it, but I deliberately holding back on it.
Recently, both l-smash and LSW have frequent and lots of changes, and that means there could be stability problem.
(as a matter of fact, one of the recently revision of LSW has a serious memory leak. The author claimed to have fixed it already, but the moral of the story is the same.)

Beware that LS/W do not have robust QC system. The main branch on GitHub is always, and only, the experimental branch. If the existing on works fine, then there is no need to make update, at least not until the new one is proven stabilized.

To All adventurers: I will not give any help, if you run into problem due to using a new version of L-Smash (Works). Use it at your own risk.

[PhotoCat]

Announcing "MotionTracking MK-II"
::
http://mavericktse.mooo.com/wordpress/archives/1950

Sorry... may I know the typical applications of Motion Tracking? Tks!

[poisondeathray]

Announcing "MotionTracking MK-II"
::
http://mavericktse.mooo.com/wordpress/archives/1950

Sorry... may I know the typical applications of Motion Tracking? Tks!

Motion tracking (in general) is used for compositing. eg. "sticking" things onto objects, as if they were in the shot in the first place . e.g .you might "cover" a blemish, you might "censor" some moving parts, you might cover up a grafitti wall , etc... It's basically a faster and (hopefuly) more accurate way as opposed to manually keyframing a layer to do the same tasks . The motion data can also used for things like stabilization

[PhotoCat]

Got it! Tks poisondeathray! I guess that's too advanced for me lol

[MaverickTse]

Got it! Tks poisondeathray! I guess that's too advanced for me lol

Yes, it is advanced indeed.
As a matter of fact, it is an active research area in "machine vision".

OpenCV (=Intel and others) recently setup a bounty asking for new alogrithms...
If any postgraduate students is reading this, you may find it attractive:

http://opencv.org/opencv-vision-challenge.html

2014-11-19
OpenCV Foundation with support from DARPA and Intel Corporation are launching a community-wide challenge to update and extend the OpenCV library with state-of-art algorithms. An award pool of $50,000 is provided to reward submitters of the best performing algorithms in the following 11 CV application areas: (1) image segmentation, (2) image registration, (3) human pose estimation, (4) SLAM, (5) multi-view stereo matching, (6) object recognition, (7) face recognition, (8) gesture recognition, (9) action recognition, (10) text recognition, (11) tracking.

[MaverickTse]

Announcing "MotionTracking MK-II"
::
http://mavericktse.mooo.com/wordpress/archives/1950

Sorry... may I know the typical applications of Motion Tracking? Tks!

Replacing someone's head with a politician's/actor's head... commonly appear on America's Funniest Home Video and other variety show.
For serious use, like what PDR said--- censorship

Another use is to generate movement track (e.g. tracking the path of a mouse over 1 hr), for investigation of movement pattern

Yet another use is "labeling". For example, you happened to shot a footage at a crime scene, when you present it to audience (the mass media do this very often), it is often desirable to highlight the victim/suspect with a colored circle, text, or similar.
Labeling also appear in sports in recent years... have a look at those football analysis(though tracking in sports field likely involve special hardware)

[PhotoCat]

Cool! Powerful stuff! Tks Maverick!

[PhotoCat]

MT-MKII... Yay, I am having fun too! It is cool!

Method 2 used. sometimes the pink ambulance got stuck under the gate...
I got luck this time...

[PhotoCat]

A practical use of background remover: when the background interfere with tracking.

With either MK-I or MK-II, the blocks are likely lost when crossing the sky-ground boundary

[Attachment 28594 - Click to enlarge]

Apply the kNN background subtraction.
Track this block-only image with your method-of-choice:
(I tracked the bottom block with MK-II mode 3, while the left one with MK-II mode 1)

[Attachment 28595 - Click to enlarge]

Use EXO Stacker to merge the two EXO, and load into Adv.Editing for final composition.

[Attachment 28596 - Click to enlarge]

I have attached the source video AVI so that you can try it yourself.

this avi source doesn't seem to work... ?!