On the fly monitoring of AR flag.

[oijonesey]

Re: Pondering how to detect advert breaks.

This is based on an assumption that there is some sort of still image at the start and end of advert breaks (i.e. the station ID/Film name bit) - can we look for sections where the image doesn't change i.e. consecutive frames can be compared and the bookmarks placed where the longest group of similar frames were found?

Otherwise is it just ITV that uses the little square of black and white scrolling bars before ad breaks? Can we look for that for the start and add time on to work out where the end is? (This length of time might need to be stored on a channel by channel basis?)

Or combine the two to spot where ads are going to start then use a frame comparison to work out exactly where the start and end would be i.e. within 5 minutes of each little B&W advert marker?

 

[Black Hole]

There are ad break detectors available on PC I believe, they have come up for discussion before (reported to have variable success rates), but anything which requires analysis of the actual video frames is going to be way too compute-intensive to be run on our Hummies. Look how long it takes the thumbnail package to extract still frames, never mind then looking at the detail of the frames to detect specific features.

I know there is a theorem in computer science which says anything that is computable can be computed by a Turing machine, but only if it has an unlimited supply of tape (working memory) and a few aeons to run (the result of the ad break removal being available for viewing by the descendants of Majikthise and Vroomfondel, truely the most interesting pundits ever to have been on the 5D chat show circuit).

 

[af123]

If bookmarks were set regardless, it would then be possible to auto-crop padding recordings to the AR points.

Wouldn't that defeat one of the points of using padding over AR? That if the AR flag would cut off the start/end of the recording then at least you have the padding.

 

[Border]

re add breaks. When you think of it, the add breaks would be the most important time for the hummy to detect changes in now/next information, and AR flag changes. Is there a change in EIT transmissions during ad breaks, e.g. different tables more frequent now next updates etc.?

 

[Black Hole]

Wouldn't that defeat one of the points of using padding over AR? That if the AR flag would cut off the start/end of the recording then at least you have the padding.

I was thinking review the bookmark positions first, move if necessary, then crop. I was a bit cavalier when I said "auto-crop". Having auto-placed bookmarks would be a good start.

 

[Wallace]

Question: I have auto shrink enabled at the root folder level. This saves me having to enable it manually. Will arbookmarks function automatically before the recording is shrunk as I realise it will not after?



Sent from my iPad using Tapatalk HD

 

[Black Hole]

Will the bookmarks still function after the recording has been shrunk??

Is there a need to increase the potential for mishap by shrinking everything? Surely it is only necessary for "keepers"?

 

[Wallace]

It was only that I was/am being lazy! I take your point but I bet I am not the only one who has chosen to use Shrink this way. After all, the option to use at the root level is there to be used!


Sent from my iPad using Tapatalk HD

 

[Border]

Re: Pondering how to detect advert breaks

On HiDef channels are the advertisements also in HiDef ? That question has probably been asked before.

Edit : (D'oh I recorded 10 minutes of a HD channel 5minutes adverts 5 minutes programs, bookmarked the middle, cropped it and realised both halves were identical byte wise or to be more technical The bitrate is constant throughout the recording. You learn something new every day!!!).

 

[af123]

Question: I have auto shrink enabled at the root folder level. This saves me having to enable it manually. Will arbookmarks function automatically before the recording is shrunk as I realise it will not after?

Yes - automatic decryption runs before automatic shrinking and the arbookmarks process is hooked into post-decrypt.

Will the bookmarks still function after the recording has been shrunk??

They should be fine. Bookmarks are stored as a number of seconds into the recording and the shrink process doesn't change that.

 

[Wallace]

Great, smashing, super...

 

[Border]

I found a couple of hours to look at this

Thank you again af123.

For some of us it takes us hours to get round the remote control!
Thank you for a very useful addition to the hummy!

 

[Ezra Pound]

As bookmarks are now used for several different functions, Arbookmarks, thumbnails setting, Nicesplice and it's original use. Do you think it would be useful to have a Web-If function to clear all bookmarks from a recording?

 

[af123]

As bookmarks are now used for several different functions, Arbookmarks, thumbnails setting, Nicesplice and it's original use. Do you think it would be useful to have a Web-If function to clear all bookmarks from a recording?

Yes, I think that would be useful.


Posted on the move; please excuse any brevity.

 

[prpr]

I think it would be useful to add and edit them as well.

 

[af123]

I think it would be useful to add and edit them as well.

It's easy enough to support editing by play-time, but not practical to do it in a visual way with thumbnails or other video content.. Is that really useful?

 

[prpr]

Given that you can't set bookmarks other than on-the-fly and therefore end up only being able to set them approximately, yes I do think it useful to be able to fine tune, even if it is just by numbers and not in any visual way.

 

[Black Hole]

I second the request to access and edit the bookmark data. I seem to remember having mentioned it before, as bookmarks can only be set in the SUI while a recording is playing at normal speed. If they could be set while paused or in slow motion play, it might be different.

If crop etc only works at I-frames, it would be useful to restrict the position of bookmarks to I-frames only, and to display (as per thumbnails) the frame at the bookmark even if only "on request".

 

[xyz321]

If crop etc only works at I-frames, it would be useful to restrict the position of bookmarks to I-frames only, and to display (as per thumbnails) the frame at the bookmark even if only "on request".

Bookmarks can only be set to a resolution of 1 second so having them coincide with an I-frame would be even more restrictive.

 

[Black Hole]

I didn't realise I-frames occur so infrequently, or that bookmarks are so restricted (no wonder I have trouble with them).

Does crop rely on I-frames? Presumably I-frames are not necessarily coincident with the bookmark quantizations, so perhaps we should have a way of fine-tuning the crop points independent of bookmarks.