Comskip Support Forum

Substantial improvements of detection accuracy usually implies slower overall speed and it seem to be about 300 fps slower compared to old method. Processing time is still very good, but maybe there is room for speed optimizations down the road?

Thanks, erik

Bug fixed. WARNING: Still very experimental Feel free to test hardware decoding in donator build of 0.81.077 and report back with your findings, in particular: - type of CPU - Type of Graphics HW - Speed with/ without HW decoding - Any detection quality differences Seems this is not for my hardware...

.dif files is still lingering in 81_077.

What are these .dif files i see in the recordings folder that seem to linger from comskip processing? They showed up when starting using the newest comskip (81_076). Is there an output option to disabled them?

Thanks, when my build environment is working again I will adapt. Erik, you should know escaping of special XML chars breaks support for the videoredo2 format (the format used for videoredo v4). Since V4 definitely are using an escape routine when opening projects putting in already escaped file nam...

There are a few more special XML chars besides the ampersand that has to be escaped: for key, val in [["&", "&"], [">", ">"], ["<", "<"], ["%", "&#37;"]] ;Escape special XML characters StringReplace, ParamEsc...

I noticed an escaping of reserved XML characters discussion are going on at the VRD site and also at the kmttg site. Since Videoredo format 3 VPrj now are correct according to the XML specifications maybe some escaping should be used by comskip as well? Maybe that also applies to format for Videored...

It works great!
Many thanks for keeping Comskip up to date, Erik!

I corrected the uppper/lower case mistakes in build 67 Can you check? Hi! Yes, thanks Erik, I can now parse it with XMLDOM :D However we're not quite there yet! Latest build has a bug where it duplicates all <CutTimeStart> and <CutTimeEnd> entries, see below. Maybe that is a consequence the <CutLis...

Thanks Erik it works! Even though dlflannery regards the faulty formatting as of little concern for his particular VAP purposes, this is the ideal moment to get the XML formatting correct. Not utilizing this moment and leave it in broken state (for MSXMLDOM purposes) would be negligence. As it is no...

So I only need to add <SceneList></SceneList> around the list of scenes? Can you edit a comskip output to how it should be and post it here? Yes! and here is an edited comskip VPrj output stripped down to the necessities that seems to work fine: <VideoReDoProject Version="3"> <Filename>D:...

Sorry erik! In 065 the scenemarkers doesnt show up in VRD5 because they are not enclosed in <SceneList></SceneList> tags. Also this VideoRedo format 3 is a mess. Hardly no line feeds which makes it harder to read and harder when scripting. <InputPIDList> no longer are used for VRD5. <CutList> has st...

Great comparison dlflannery! Erik and dlflannery, what do you think about the idea to rename version 1, 2, 3 videoredo VPrj formats. It's quite confusing and frankly who does really keep track of which format version is which, besides very initiated people? I think it would be more logic in renaming...

Is the v3 output wrong? or are the scene markers missing? Can you send me some annotated examples of the v5 format? Hi erik! Yes, when using comskip with videredo3 format in .ini no scenemarkers are found at all. Im a bit unsure on the videoredo5 format, but when using the Ad-scan feature in v5 the...