Kodi Community Forum

Full Version: High-Definition Video Flagging (library filtering/sorting & skin thumbnails overlay)?
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
The patch I said I'd post on Friday I just put up today. Stupid work doing a redesign of a feature after a release candidate + costume party weekend = Zzzzz.
great! it isn't already in svn, is it? can't find it at trac
i for one wouldn't announce my sucking at basic searching that loud. it's sitting rite in the original ticket
aha, that's about it, original ticket... i was looking through all new tickets, silly me.
jmarshall Wrote:Some stuff is committed. I have some queries about VideoInfoTag changes. What are these for? It seems to me better to just stick with one tag name to save confusion?

Also, I'm going to change the aspect ratio stuff to work on ranges rather than the current system. Reason is that the current things either overlap, or have spaces between them - thus some fall through the cracks. Better IMO to use something like the geometric mean of each ratio as a cutoff.
The VideoInfoTag, the codec/codecid change reflects the decision that fourcc should be used instead of decoder codec. I know the nfo generator can make this decision itself, but there are already utilities out there that put both out so I thought this would be acceptable. The displayaspectratio thing was solely for compatibility but not as tied to that one.

And you're right about the aspect ratio (VideoAspectToAspectDescription), it has both overlap and spaces between the range! I originally wrote it as "find the closest" but then I considered that now I was just labeling things for the sake of labeling them. For example if we calculate the aspect to be 2.00, what is that? It could actually be an old 2.00 SuperScope video but "closest" would label it 1.85 which is probably wrong. Then I tried a fixed delta method which had too much overlap, and compromised with the epsilon scaling method. I welcome a better heuristic but do think there should be gaps because some things are just too far away to be called.
In the case that things are too far away, currently we just return nothing at all, right?

IMO the lesser evil is to show the closest in that case.

The codec vs codecid I'll add, as I can see codec being more useful to be quite honest than fourcc, but let's stick to one tag for the aspect.

Cheers,
Jonathan
jmarshall Wrote:In the case that things are too far away, currently we just return nothing at all, right?
Yeah that's right. I thought that would be preferred to prevent it from being flat out wrong, but the more I think about it the less I think it is an issue. Closest will probably work out ok.

I agree with you that knowing the decoder is more useful than knowing the fourcc, but there was like a 2 page discussion about it a few pages back and it seemed that there were 2 groups of people. Some didn't care about the codec at all, and the others wanted fourcc. So the probe looks at fourcc first. I figured I'd emulate that in the nfo reader but really not a big deal if it does it or not.

<apsect> you're right about though. No use making a this or this or this or this for the same value.
Let's leave it at <codec> and we can decide what we want to fill it with (i.e. internally from XBMC) ourselves. IMO some normalisation could go a long way here, to save skinners from going insane Smile

Will fix up the aspect stuff later today with any luck.

Thanks again,
Jonathan
What are the plans regarding a real integration of this amazing work to SVN ? As far as I know, we still have to apply the patch, right ? What's been blocking you, Capn ? Wink
you are full of bad information
bertrand Wrote:What are the plans regarding a real integration of this amazing work to SVN ? As far as I know, we still have to apply the patch, right ? What's been blocking you, Capn ? Wink

What spiff means is that it's been in SVN for a while now. Perhaps you should read earlier in this thread?
I read this thread as much as I can, but it's honestly becoming hard to extract the actual, definitive info out of a 57 pages thread, right ? So I apologize for asking a bad question.

Edit: it was not THAT long ago, not even 3 weeks for the main branch Wink
The flagging code went in around June 15th, but the filter code just went in on... Friday night I think. jmarshall and I are just discussing a couple potential minor nfo importer changes, and possibly changing what's displayed for some aspects that are currently blank.

EDIT: Shows how many people use the smartplaylist filter, I thought for sure there would be 100 posts in here saying "It doesn't woooooooork IMO" where IMO = the way only I want it to Image
CapnBry Wrote:Shows how many people use the smartplaylist filter, I thought for sure there would be 100 posts in here saying "It doesn't woooooooork IMO" where IMO = the way only I want it to Image
The majority of XBMC users are probably still more likely to wait for the next stable point release (which should be this November) before updating from the 'old' XBMC 9.04 stable point release, so I am sure A LOT more feedback will come then, and we will by then hopefully also see many more skins taking advantage of these new features.

Wink
Finally compiled XBMC from SVN with the hdflagging changes, and absolutely no issue so far (using the MediaStream files posted above). Tried it on multiple HD or SD TV Shows and movies.

I still need to dig the thread out to be sure of when files are scanned... :p
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43