Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jmone]

Thanks (V30.0.66):
- Matt, new Tag to View all (including Read Only) is working well
- Hendrik, Scan Type also works well.

[jmone]

Couple of other areas in MC that needs an update to show the new Audio Info (eg ATMOS) is:
- Audio Path
- Stream Selection

[Hendrik]

The latest build now splits black bar detection out from the other information gathering, and also gives you an option to run it during import akin to audio analysis. Its still running as a separate process and is possibly prone to a timeout (even after increasing the allotted time), but maybe I should just remove the timeout there and make sure the canceling works.

[zybex]

*decreased ?

Thanks Hendrik, good options

[Hendrik]

The timeout was increased, so it has more time to complete.

[Hendrik]

The next build will also support Blu-ray directory structures (BDMV) but DVD structure rips (IFO) will take a bit longer.  We don't have as much support code ready for it.

I'm also working on bringing "immersive" audio codec detection to LAV, which is required to make it show up in the Audio Path, and help with more dynamic bitstreaming options.  That should be coming soon.

[lello]

The timeout was increased, so it has more time to complete.

Audio analysis takes much more time

[eve]

The next build will also support Blu-ray directory structures (BDMV) but DVD structure rips (IFO) will take a bit longer.  We don't have as much support code ready for it.

I'm also working on bringing "immersive" audio codec detection to LAV, which is required to make it show up in the Audio Path, and help with more dynamic bitstreaming options.  That should be coming soon.

Dude, you're killin it. Thanks! Glad to see "Immersive" codec detection coming to LAV!

[Ashfall]

This is really great!  I previously used a custom tag to manually fill video scan and HDR type for imported videos.  Really appreciate this.

I've noticed a few videos don't get the video scan field populated.  The only thing they seem to have in common is that they are 30 seconds.  Longer or shorter videos work fine.

Update:  after updating to 30.0.73 this is working.

[eve]

Is there any chance to have this populate for items that aren't in the JRiver library when a file gets added to the play queue?
It already reads the corresponding .nfo as it should to pull its metadata, and a thumbnail from a frame grab seems to get generated by JRiver.

Thanks

[jmone]

More testing, and it's looking good overall

The scan does not work on MPLS (created by Particles). 

[elprice7345]

Great new feature!

I was playing around with some of the video data fields and noticed an anomaly.

MC shows the frame rate as 120 for one of my videos, but MediaInfo detects it as 59.94.

Is this a bug or an issue with my setup?

[jmone]

Testing on 30.0.75 : Ran both Analyse Video and Video Black Bar Analysis over all my Videos (6,779)

Analyse Video:  Looking really solid,
- (currently) Not supported for JTV, BDMV3D, IFO
- Works now with MPLS Particles
- Only other ones that did not get any info were 2 x UHD BDMV Movie rips, and (for some reason) I have 25 x BDMV (not MPLS) particles that also did not work (but I'll re-create these as MPLS).  No idea why the 2 x BDMV did not get any info (BD UHD of "Fury" and "The Patriot")

Black Bar Analysis:  This worked OK - and we all know this will never be "perfect", but some #'s and thoughts:
- 12.5% did not get any Video Crop values
- A significant set of others are just wrong
- When running Black Bar Analysis, it seems to overwrite the Aspect Ratio BUT only if it had a value before, it did overwrite those that had "0" in it after doing the prior Analyse Video

"Aspect Ratio" Field:  It seems this field can be updated with various values pending what is run last:
- Update Library from Tags = uses "Dimension" Field
- Black Bar Analysis = use "Video Crop" Field
- I've also got some 500 videos where the "Aspect Ratio" was set to "0.00" at some point and it seems to be "stuck" on this value and neither running Update Library from Tags or Black Bar Analysis will change it.

I'm not sure what other think???, but my suggestion would be:
- Use the "Dimensions" field to calculate "Aspect Ratio" as was traditionally done
- Use the "Video Crop" field to populate a new "Aspect Ratio (Crop)" field
- Make the new "Aspect Ratio (Crop)" field user editable so the dodgy ones can be manually fixed (note: I see the "Video Crop" & "Aspect Ratio" fields are user editable which is good)

Note: Zybex came up with a calculated "Aspect Ratio (Crop)" for SOT so I'd not want to cut across his bow

[Hendrik]

I also have a few files that report Aspect Ratio of 0.0, I'll be looking at them to see whats up with that. This seems to be the value the analysis calculates, but i'm not sure why yet.

I do not plan to have a separate field at this time. I don't see the value in knowing the distinct aspect ratio with black bars or without, as we really only care about the active image aspect ratio - its also not used in playback and is purely informational. (feel free to present an argument why it would be useful!)
The logic should generally try to avoid resetting it when you have a value obtained from the black bar analysis.

As for cropping being wrong -
All my files analysed fine with .75 now, none were skipped any longer.

If you have any files that were skipped, or analysed to obviously wrong values, please do share.

[lepa]

I have some SD 4:3 (dimension 720x576) television recordings inside matroska which gets detected like
Video Crop 0x0x720x576
Aspect Ratio 1.00

Also some with dimension 1800 x 1080
Video Crop 0x0x1800x1080
Aspect Ratio 0.00

[Hendrik]

The 0.0 case should be fixed in the next build.

If it actually has a value, are you sure its not actually playing as a square video as well? Something in the file would have it compute that.

[lepa]

'Video.DisplayAspectRatio_String' from Mediainfo gives me 4:3 and file also plays as 1.33

[jmone]

I'm certainly not going to die on a hill about aspect ratios (as you say they are (currently) informational)....and I'm a 16:9 display only user.... but my argument would be:
- One is about the video frame
- The other is about what part of the video frame the content uses.

It may make a difference for those with the screens that have masking?  No Idea....

It also just feels wrong, to have one field that can have two different values pending what process was used when.

I've got all sorts of odd looking aspect ratios (but I'll send you a link to an XL dump of my videos with the new fields) that will show the ones skipped, the 0.00 and the ones that are just... odd (you can ignore the test pattern ones).  TBH I'll personally just stick to the calculated ones over black bar detection with my setup, but thanks for looking.

[Hendrik]

I can't really do anything with a list of metadata. A few select samples would be helpful. (Also, Aspect Ratio seems to not be in that?). Don't need 0.00 samples, thats just a bug and will be fixed.

In general, its not making up data for aspect ratio, so if its coming up with a value that is not 0.00, then that is based on data from the file. In the simplest case just w / h, but many files also have an aspect ratio beyond that for anamorphic rendering. So if you have a file with the dimensions of 720x576, which has a natural aspect of 1.25 (w / h), then something is transforming it to 1.00, that value is not an accident.

Sometimes there can be conflicting info in a file. The video stream says one thing, the container another, and which value to give priority is a long debate. I'll double check the preference matches what playback uses.

A sample would of course let me make sure.

[jmone]

Fair enough (I send the fixed XL file with aspect ratio this time )  I'll redo once the 0.00 thing is fixed and report back with specific examples but two I mentioned earlier that don't get any values are the UHD BD version of "Fury" and "The Patriot"

[Hendrik]

For weeding out weird Black Bar detection things, I've setup this smartlist for myself:

Code: [Select]

[Media Type]=[Video] -[Video Crop]=[] -[Video Crop]=[0x" -[Video Crop]="x0x" -[Video Crop]=[2x" -[Video Crop]="x2x"
Basically what it does is find files which don't have one of the borders at zero (or at 2 pixels) - which would indicate typical cropping behavior - eg. just one side having letterboxing/pillarboxing.
If both sides have significant cropping, either the video is weird (encoded bars on both sides, it happens), or something is up.

No guarantees that other files which seem to have reasonable values won't be wrong in other means, but these are the most obvious cases.

Personally, my actual video library is pretty high quality files only. But I have a ton of sample files I'll have to throw into my test library as well over the next few days.
I do see a few outliers in my own library that I'll be looking at as well.

Once I have nailed down the issues with my files, if any issues for you remain, a sample would indeed be helpful.

[jmone]

Nice - that smartlist did show a bunch of questionable vids (home vids, test files etc), but also had 100+ BDMV/MPLS rips.

[jmone]

eg: Terminator 2: Judgement Day UHD BD rip - Video Crop is 408x1000x3758x1312

[lello]

For what it's worth, today I wanted to analyze my library and out of 219 files the Video Crop field is populated only in 8 cases.

[zybex]

Subtitle Language is not parsing unicode - that's Norwegian Bokmål

[Hendrik]

I've fixed some issues with black bar detection, made the detection window a bit longer, and my samples all seem to be improving.
We already sample 7 different spots within the first 30 minutes of a video, and reading around 5-10 seconds at those spots to get a good idea of the video. Other then parsing the full video I'm not sure where there is a diminishing returns in expanding on how many spots we check to give the biggest coverage.

Subtitle Language is not parsing unicode - that's Norwegian Bokmål

Actually the language was encoded wrong in our list from language code to language name, but fixed regardless!

[zybex]

Maybe spread out those 7 samples to the entire length of the movie? interval=length/8 or something.

Kursk (2018) has the first 17 minutes at 1.66 and then opens to 2.39 until the last 2 or 3 minutes, which change back to 1.66. MC sets the Video Crop to 0x0x1920x804, which seems correct.

[Hendrik]

I changed it to spread out a bit more over the video depending on duration. And yes, the crop should cover the "biggest" possible active image. So if it changes mid-video, it would ensure both (or more) options will be fully shown inside the crop rect.

[lepa]

With latest beta I haven't found any issues anymore with crop detection. Good work!

[jmone]

I'm do in order a full:
- Update Library from tags, then
- Analyse Video
- Black Bar detection

It will take hours!

[Hendrik]

Update Library from Tags includes the same processing Analyse Video does, no need to do it separately. Analyse Video just does less, eg. not overwriting actual tags in your DB etc.

[jmone]

Thanks - makes sense and just saved me some hours (up to 881 of 6778)!

One other thing I found is I had a live URL in the Video section (for a Camera I think - now deleted the entry).  When Analyse Video hits that it stop and you can't cancel it, you have to close down MC and then kill the worker from Task Manager (or reboot).

[OverTheAir]

... but two I mentioned earlier that don't get any values are the UHD BD version of "Fury" and "The Patriot"

Probably irrelevant but both those discs had AACS 2.1 and IIRC very few other discs have ever used it. Can't see why that would make a difference vs behavior with AACS 2.0 protected discs, regardless of whether protection has already been removed at time of rip or if its done dynamically on the rip at time of playback, but thought I'd mention it.

[jmone]

Yeah interesting point.  Both have been decrypted but I wonder.  Lets see what happens after the latest set of updates I'm doing.

[jmone]

Testing on 30.0.76 : Update Tags over all my Videos (6,779)

Looking really solid and I'm personally very satisfied with the results.
- (currently) Not supported for JTV, IFO, ISO
- Works now with BDMV3D and MPLS / BDMV Particles
- Only failure was on the 2 x AACS 2.1 BD UHD rips ("Fury" and "The Patriot") - Weird, I might try to re-rip these.
- The 0.00 Aspect Ratio is fixed (well It's gone from 500+ to 4 and these are all WMV and are not what you would call great quality)
- Aspect Ratios look good at this stage

Video Bitrate: About 2/3rds don't get a Video Bitrate.  Having a look at these in MediaInfo, I see that many just don't have a listing for Bitrate at all, but lots have the following format if that is of interest:

Code: [Select]

Bit rate mode : Variable
Bit rate : 19.7 Mb/s
Maximum bit rate : 35.0 Mb/s
Off to start Black Bar detection

[jmone]

Black Bar detection is still grinding on but getting there.

Re-ripped Fury and The Patriot with makeMKV and these analyse just fine (must be some DeUHD thing back in the day on 2.1 discs). 

[jmone]

So after 8 odd hours Detect Black Bars finished.  Looking really really solid!
- (currently) Not supported for JTV, IFO, ISO, bdmv3d
- Minimum Length of Video seems to be 45sec (no probs with that)
- Used Hendrik's Smartlist to look at potentially odd crops but these were only 2% of my files and they were mostly actually correct with some weird matte (combo of letter and pillar box) or older analogue transfers without crisp edges. 

No Video Crop Calculated.  Not many of these, but include:
- Taken (BD FHD)
- Sherlock Holmes (BD FHD)
- Pink Floyd Wish You Were Here (BD FHD)
- Board Walk Empire (BD FHD) S4 D1,2,4  S5 D3

Exceptions:
- Grand Budapest Hotel (BD FHD):  Reported as 272x22x1644x1058 : One of those movies that likes to switch between mattes

Notes: These are examples of some poor mastering that is shown up by this new feature
- Patriot Games, Start Trek 5 & 7, Rambo 2, and Shaun of the Dead (all BD FHD) that have a bit of unnecessary black bar edging on one side

The only thing I've found is if I clear the "Video Crop", the Aspect Ratio is not recalculated from Dimensions.  I have to run "analyze video" again.  Not a biggie but worth mentioning.

Well done!

[Hendrik]

No Video Crop Calculated.  Not many of these, but include:
- Taken (BD FHD)
- Sherlock Holmes (BD FHD)
- Pink Floyd Wish You Were Here (BD FHD)
- Board Walk Empire (BD FHD) S4 D1,2,4  S5 D3

What file types are those? BDMV?

With BDMV it only picks one m2ts file to look at, I wonder how solid that is on fancy discs with lots of branching, or other shenanigans. But I can't really justify having a full BD demuxer there at this point. Maybe once I've made one for Linux, it can be used here..

[jmone]

All are BDMV (rips from original BD):
- Taken : One big M2TS, does not look special in any way
- Sherlock Holmes 2 : Yup, it is a seamless branding disk with several M2TS.  Largest M2TS is a 30min segment from the middle of the movie and should produce the correct crop
- Pink Floyd Wish You Were Here : Forget it, it's one of those Audio BD with a few Video segments, but the main segment is Audio only with Black Video
- Board Walk Empire : It's the typical TV BD with one large M2TS per eps.  Can't see anything obviously different between the Discs that work fine in this series and the ones that don't

It's really not worth tracking down.  At best it is a few discs out of bucket loads.  Just reporting in.

[lello]

With bdmv the audio title field is not populated: is this normal?

[Hendrik]

Blu-ray discs hold no title information for their streams. You get language and codec in other fields.