@Sergey Sharybin (sergey) would be better to decide what to do with this.

This means, that there is something wrong with your user preferences.

In T60460#616922, @Andrew Sprok (sprad001) wrote:

Here is the attached result. The LibreOffice file should open in Word as well.

Plaintext is preferabe :)

@Andrew Sprok (sprad001)
Can you please confirm my earlier comment, that I am following the steps required to reproduce crash?

Haven't realized, that it is possible to add image sequence...

Remove unused variable

This function argument swapping seems rather ugly to me, but I didn't want to modify original data...

Can not update diff to add selection info label.
So here's the code I would add.
I will accept this anyway, will see what other devs will say about this, since it's not critical.

Fix comment style
Remove unused variable

This panel is included in VSE preview.
If ibuf->metadata is NULL it crashes Blender.

In 2.79 checkerboard is not drawn in backdrop.
So it will be disabled.

Only now I realized, that this is 2.79 bug report
Can not reproduce this in 2.79 f4dc9f9d68b or in 2.8

Please correct me if I am doing this wrong:

• create new image sequence 0.jpg - 100.jpg
• click on panel->srtip->change data files
• select 0.jpg - 300.jpg
• change strip->length to 300
• click render->render animation

Ok, so I can confirm this. There are 2 issues:

Sorry it tried to embed movie, but I can't display it... I see it now, will test again.

@Sebastian Parborg (zeddb) That shouldn't be the case.

In D3934#97372, @Guillaume Matheron (mathers) wrote:

I read your updated doc, and here are a few notes :

• I think we can keep a single thread for prefetching : rendering a single frame should already be multithreaded in many cases (renders, effects, etc...). Would doing this remove the need for error-prone scene copying, or is it inevitable ?

Finish refactoring draw_image_seq

Updated summary to short-ish documentation.

Got that later :)

In T61169#612940, @William Reynish (billreynish) wrote:

Actually, the bug seems to have nothing to do with the Opacity property, but with using Alt-I. If you use Alt-I to remove keyframes, something isn't being updated correctly.

The problem seems to happen on *all* properties if you use Alt-I to remove keys.

I was chasing something similar... Sequencer doesn't signal to depsgraph to update data and animdata are evaluated in depsgraph.

In T61043#611875, @Joerg Mueller (nexyon) wrote:

The bug was introduced in rBdbe23c545649, so it only exists in 2.8, not 2.7. And yes @Richard Antalik (ISS) I can confirm that, but it has nothing to do with this bug.

In D4288#97146, @Guillaume Matheron (mathers) wrote:

Which repo ? sobotka/blender on branch vse-cache ?

In D4288#97115, @Guillaume Matheron (mathers) wrote:

In D4288#97063, @Richard Antalik (ISS) wrote:

By doing incremental changes, this situation can happen if we are not extremely careful. That's why in case of big rewrites we work on branches that if functional and "bug-free", they will be merged in master branch.

Is there a branch that I can checkout in order to test D3934 and review individual commits ?

My patch is not outdated it should apply cleanly and partial review is still pretty much possible.
You can try to rewrite it from scratch - I have no power to stop you.

@Joerg Mueller (nexyon), not sure if you are involved in audio codec stuff, can you confirm my claim: "especially mp3 can report longer length of data stream then actual length." This is just a general observation from number of mp3 files, that behave this way.

I can confirm a frequent crash due to drawing sound waveform when strip has endstill frames. drawseqwave seems to be using length of strip to calculate data length

Updated diff.

@Christopher Anderssarian (Christopher_Anderssarian) was right... There's lack of communication.
Sorry, I can't keep up and I am already paying more attention to Blender then to my work :) Hope, that my boss won't read this...

In T61056#610933, @Sebastian Parborg (zeddb) wrote:

@Richard Antalik (ISS) have you reproduced this or did you simply just tag it?

Both :)
I haven't looked yet at whole color transform pipeline in sequencer, but I didn't think, that this is bug.

Slow down a bit :)

Thank you for reporting and investigating this.

Ok So Brecht said, that this(default settings) will be dealt with.
He also said, that look Base Contrast should not affect the output, which seems to be correct. So it was assumed, that no transformation is taking place. However this seems to not be correct and either transformation with unity coefficient is executed, or imbuf is converted to float, or something else is going on.

Can you please run blender from command line with --debug argument and post output here? (or rather paste it there http://pasteall.org/ and link it here)
If it will produce a crash report it would be nice to include that also.

Is the source of slow rendering known?
I should at least look at this, as I don't use blender for modelling therefore I am less likely to notice and trace such issues

To be sure about _exact_ numbers I did a comparison between 2.79 and 2.80:

Now I am thinking that good question is why is this "bad output" displayed "correctly" in blender

It wasn't clear from description, if this applies to renders from video sequencer(VSE bug) or all renders(FFMPEG bug).

There are some misconceptions here.

In T60912#608438, @Guillaume Matheron (mathers) wrote:

• Disk cache files are image sequences generated from the output of the sequencer (3). This cache is initially stored in RAM, then dumped to disk when the memory limit is exceeded. These images are the result of the full composition (4), and are invalidated when a strip is moved, or anything is changed in the sequencer. This allows for real-time preview and scrubbing, even in scenes with heavy compositing. Pre-fetching is used.

In T60912#608235, @paul szajner (szap) wrote:

Hello,
I'm not sure to understand , do you plan to remove proxy made of image sequence ?

Not now, not until equivalent or better system will be in place
Playback of images should be fast without proxies. Proxies may make sense to be used for 4K or bigger resolution Image sequences(sets of images instead movie file) with a lot of processing, or some demanding formats.

In D3597#96377, @Christopher Anderssarian (Christopher_Anderssarian) wrote:

A design task clearly detailing the goals, timeframes, scopes and functionality would be desirable

Got this.
I intended to create some roadmap, I can define goals and abstract functionality.
Currently I should be working on prefetching, with aim to get it up and running in 2.8 release, which should be doable as code is already working.
I delved into a lot of stuff lately, maybe unnecessarily, well...

I am not absolutely sure, if I am happy with splitting of proxy panel.
Will apply this patch and try to work with it and think about it a bit more.

In D3597#96364, @Guillaume Matheron (mathers) wrote:

I'm not sure why we need to change this API ? I'll take a deeper look at your patch.

What can be done to proxies now is

• cleanup in sequencer part
• PNG codec support was not a bad idea. It has a bit worse performance, but it has transparency.
• Even more demanding lossless codec can be implemented - FFv1, huffyuv...
• it may be useful to introduce scale to specific size - VSE should support this as-is I think. Also See last section.

If I understand correctly, the plan is to :

• Drop proxy support for all sequences except movies (it would also be dropped for clips and image sequences right ?). All proxies would be ffmpeg-encoded avi files, no image sequences.

Not quite. Movieclips do have "own" proxy management. Builder is shared, but you can not build proxies for movieclip from sequencer

In D3597#96209, @Guillaume Matheron (mathers) wrote:

You would have to look by date, but this patch needs to be reworked:

• to keep things simple, code cleanup done in this patch should be dropped

You mean this patch should be re-written from scratch ?

In D4199#96184, @Campbell Barton (campbellbarton) wrote:

Currently this patch is getting too big to properly review.

True.

In D3597#96185, @Guillaume Matheron (mathers) wrote:

Hi @Richard Antalik (ISS) , I'm trying to test your patch, but I am unable to apply it to the master branch. What is the base commit for this patch ?

Once I can build it, I think I can tidy this patch up a bit, remove warnings etc.

seems similar to T59926
Haven't realized, this may be only present in 32bit build. i switched to 64, haven't noticed bug anynore and closed it.
I can still reproduce also T59927 on 32-bit

Oh I forgot to link to bug report in earlier comment.
Also T56755 looks similar to this.

In D4219#95901, @Sergey Sharybin (sergey) wrote:

Doing such a manual checks should only be the last resort. I would expect that passing clip->flag BKE_movieclip_get_ibuf_flag() will solve the issue.

The patch forces proxies to be used even for clips which are configured to not use proxies. Which is ok-ish, for until there are outdated proxies left on the disk, and clip has proxies disabled.

There should be no problems with this backport..
@Brecht Van Lommel (brecht) If you decide this should be done, can you please do merge? I am not feeling comfortable doing this on official repos...

Add fallback full size render, if proxy not built

In D4199#95850, @Peter Fog (tintwotin) wrote:

I'm must confess that I'm not ready to give up on this yet...

Added tag for easier tracking

I could not resist to do a little experiment of jumpy panel: https://photos.app.goo.gl/dXQyg23Z5sTzJUVH6

To reproduce please follow T54117

I was looking at menu items,
Biggest problem are operators - many of them can be run with properties, that would contradict their description, or depends on selected sequences, etc...

In D4199#95618, @Peter Fog (tintwotin) wrote:

In D4199#95594, @Richard Antalik (ISS) wrote:

• SEQUENCER_OT_Move
  • What about moving to nearest snap points(another strip start, end, markers, channels etc) in desired direcion? I got this coded, can move large blocks, but need better collision handling(need this anyway, so why not...)

Ctrl+drag playhead/strip will snap to next cut, so snapping with Ctrl pressed would be in consistency, but the default move should be without snapping.

In D4199#95561, @Peter Fog (tintwotin) wrote:

@Richard Antalik (ISS) Thank you for investing so much time in this. The "space_sequencer.py" doesn't seem to be updated to reflect the changes in "sequencer.py"? When I replace these two files in a fresh build of Blender the entire UI is broken. So I can't test your changes.

In D4199#95531, @Campbell Barton (campbellbarton) wrote:

@Richard Antalik (ISS), could you identify any of the changes here that'd be better done in C? (as TODO's in comments for eg).

In the case of extending "Select Grouped" this is something you could probably do even without much experience in C.
We can help you get started compiling Blender.

Otherwise, someone else can add this.

@Peter Fog (tintwotin) I should say, that I am modifying these scripts, so please don't make very big changes.

In D4199#95342, @Peter Fog (tintwotin) wrote:

I think it is better to wait with the keymap until sequencer.py and space_sequencer.py are good to go, as the keymap may change with changes in the other two files. NB. though @Brecht Van Lommel (brecht) has invested a lot of time in evaluating the code, there still seems to be a good deal which hasn't been commented(and I fear that it isn't because the code is perfect). Maybe Brecht can let you know what still needs a first evaluation? It is not all remarks I'm skilled to solve myself. And as mentioned in another comment here, there are several open design questions, which someone with the right authority needs to decide on.

Can not reproduce even in 2.79
Maybe can you save blend file, with which This can be reproduced?

Tried to reproduce, however in output settings value of Video Codec item is empty.
What codec do you have set?

Updated patch based on latest version

Hello, sorry for lack of communication.

Fixed by 3610f1fc43d0

@Christopher Anderssarian (Christopher_Anderssarian) disabling V-sync did solve issue, that refresh rate depends on open panels.

@Brecht Van Lommel (brecht) I will review and modify this. It was submitted as-is, because I wanted to comunicate stated questions with authors.

During debugging in MSVC (probably not up to date version), debugger was confused about Sequence type.