Screenshots

This article is aimed at providing useful information regarding adding screenshots to your description and the software used.

Adding screenshots to your description is necessary, as it's meant to preview the video and subtitle quality for the torrent. For that reason when taking screenshots they must be in source resolution, PNG (lossless) format, unless the video is anamorphic in which case they need to be in display resolution.

r2 - 2013-07-04

Taking Screenshots

1. Recommended: AvsPmod (& AviSynth, Dither)

   Most flexible solution, but requires a few steps to set up.

   • + using Firesledge's script provided in this thread pretty much guarantees the screenshots to be the most faithful representation of the source
   • + high quality scaling for comparing releases with different frame dimensions
   • + AvsPmod provides efficient and convenient ways to compare multiple releases and batch export screenshots
   • - rendering subtitles requires them to be demuxed from the release

2. Good: MPC-HC + madVr

   Easy to set up solution that doubles as THE high quality video playback solution on Windows, but not as convenient as method #1

   • + video quality on par with the AviSynth method
   • + easy to set up (install CCCP) and use (like a media player), doesn't require any demuxing or indexing
   • - does not supporting scaled screenshots for comparing releases with different frame dimensions
   • - not as well suited (or intended) as AvsPmod for more extensive comparisons, comparing releases side-by-side possible but cumbersome

3. Good: mpv (opengl-hq)

   Cross-platform solution also available on Mac OS X and Linux, not as convenient as method #1

   • + easy to set up and use if you're familar with the command line
   • + access to advanced configuration options and quite powerful video filters for a video player
   • - video quality very slightly inferior to methods #1 and #2
   • - not as well suited (or intended) as AvsPmod for more extensive comparisons, comparing releases side-by-side possible but cumbersome

4. Avoid: MPC-HC + EVR CP

   Default setup when installing CCCP or MPC-HC/BE.

   • ! not a fault of EVR in and by itself, but an all too common error source: uses video "enhancement" features of your video card driver which actually destroys the video (and your screenshots) in several ways, including but not limited to:
     • fucking up colors,
     • adding ugly sharpening,
     • creating sugoi banding, and
     • other artifacts.

   If you absolutely have to use the EVR renderer for taking screenshots, make sure all video enhancement options of your graphics card driver are turned off

   • - possible pixel shift to the left (tested on recent NVidia drivers)
   • - slightly inferior video quality compared to madVr (defaults to bilinear chroma upscaling, random dithering using lav video decoder)

5. Do not EVER use: VLC

   Atrocious video player with terrible video quality

   • - uses wrong matrix for RGB conversion (results in wrong colors)
   • - uses point upscaling for chroma planes
   • - introduces strong banding
   • - all the other bugs (including some that haven't been fixed in years) make it equally unsuitable media player. Seriously, stop using VLC!

6. Do not use until further notice: FFmpeg

   Versatile command line video/audio processing tool unsuitable for taking screenshots

   • - incorrectly uses BT.601 matrix for all content (results in wrong colors)
   • - colormatrix filter causes strong banding
   • - uses point upscaling for chroma planes and ordered dithering (low video quality)
   • - not practical doing screenshot comparisons (as a standalone tool), doesn't seem to be able to take frame numbers as seek points

AvsPmod

Required Components

Software	Version
AviSynth	2.6a4	Download
AvsPmod	2.5.1	Download
FFMS2	r743 (10bithack)	Download
Dither Avisynth Tools	1.22.1	Download
Masktools	2.0a48 for 2.6a4 (06_taro)	Download
xy-vsfilter	3.0.0.214	Download
FFInfoEx	r1	Download

This tutorial also assumes you're running a fully updated Windows 7/8 machine with all recent VC++ distributables, .net Frameworks and whatnot. Things probably also work on Vista/XP and you might even get them to run with Wine on Linux, but you're on your own.

Setting thing up the environment

1. Install AviSynth
2. Open your AviSynth plugins directory in Explorer
   Unless you changed the the AviSynth install location, your plugins directory should be
   (enter these into your explorer location bar):
   • [On 64-bit Windows] %ProgramFiles(x86)%\AviSynth 2.5\plugins
   • [On 32-bit Windows] %ProgramFiles%\AviSynth 2.5\plugins
   There should be only 3 files inside the directory: colors_rgb.avsi, DirectShowSource.dll, TCPDeliver.dll
3. Open the Dither Avisynth Tools archive (dither-1.22.1.zip) and unpack the following files to your AviSynth plugin directory:
   • avstp.dll
   • dither.avsi
   • dither.dll
   • mt_xxpand_multi.avsi
   The docs folder inside the archive contains detailed documentation on the dither tools, so you might want to extract this to your AviSynth Docs directory or anywhere else you'll find it in case you need it later
4. Open the downloaded ffms2 archive and extract the contents into your Avisynth plugin directory
   • ffms2.dll
   • ffmsindex.exe
5. Open the downloaded masktools archive and extract mt_masktools-26.dll to your AviSynth plugin directory
6. Open the downloaded AvsPmod archive and extract the AvsPmod folder to:
   • On 64-bit Windows: %ProgramFiles(x86)%
   • On 32-bit Windows: %ProgramFiles%
   Create a shortcut to avspmod.exe, or pin it to the taskbar/start menu
7. Copy the downloaded VSFilter.dll and FFInfoEx.avsi to your Avisynth plugin directory

Getting started with screenshot comparisons

1. Start AvsPmod and open the settings dialog (Options > Program settings)
   1. On the User Sliders tab check Hide slider window by default (The slider window usually only gets in the way)
   2. If you are using more than one screen, it's usually wise to check the Seperate preview window option, so you can have one screen dedicated to the video and one to the script
   3. On the Misc tab, uncheck the Show warning for bad plugin naming at startup option
2. Go to Options > Snippets and click on Insert. Insert the following script template as snippet and give it a tag you will remember
   (Script by Firesledge with a few additions)
vidfile="VIDEOFILE" subfile="" c1 = FFVideoSource (vidfile, enable10bithack=false) c2 = FFVideoSource (vidfile, enable10bithack=true, varprefix="c1_") w = c1.Width () h = c1.Height () wm4 = (w + 3) / 4 * 4 hm4 = (h + 3) / 4 * 4 b10 = (c2.Height () > c1.Height ()) target_width = w target_height = h display_timecode = false matrix = (c1_FFCOLOR_SPACE == 1) ? "709" \  : (c1_FFCOLOR_SPACE >= 5) ? "601" : Dither_undef () par = (c1_FFSAR_NUM > 0 && c1_FFSAR_DEN > 0) \  ? Float (c1_FFSAR_NUM) / Float (c1_FFSAR_DEN) \  : Dither_undef () tv_range = (c1_FFCOLOR_RANGE != 2) (b10) ? c2 : c1 (VersionNumber () >= 2.6) ? last \ : (b10) ? Dither_addborders16 (0, 0, wm4 - w, hm4 - h) \  : AddBorders (0, 0, wm4 - w, hm4 - h) (target_width != c1.Width() || target_height != c1.Height()) ? Dither_srgb_display (lsb_in=b10, matrix=matrix, par=par, tv_range=tv_range, w=target_width, h=target_height) \ : Dither_srgb_display (lsb_in=b10, matrix=matrix, par=par, tv_range=tv_range) (StrLen(subfile) > 0)  ? TextSub(subfile) : nop() FFInfoEx(vfrtime=display_timecode, cfrtime=false, frametype=false, x=last.width-6, y=4, cmt="GROUPNAME", align=9, shorttime=true)
3. Now that you're all set, let's load the videos to compare.
   1. Click into the empty script window and hit F7 to get a list of your snippets. Since there's only the one snipped you just created, hit TAB to insert it
   2. Replace VIDEOFILE in the first line with the full path to an episode of the first release you want to compare and GROUPNAME in the last line with the appropriate group tag
   3. Hit F5 to load (or refresh) the video. Loading a file for the first time might take a few seconds, since the file is being indexed first.
   4. Save the script
   5. Hit Ctrl+N or double click next to the current tab to open a new script tab
   6. Repeat from Step 3a
4. Syncing up the releases (easymode)
   1. Check if the video of all releases matches up, by going through them with Alt+1, Alt+2, Alt+3, ...
      The best way to do is is to seek to the last frame before a scene change, check if it matches up across tabs, then seek to the next frame and check again
      (Seek by using the arrow keys (left/right: 1 frame; up/down: 1 second) or clicking the timeline for larger time distances)
      
      Note: You may find out that not all subsequent frames match up or worse, the releases running out of sync over time. If it's only single frames, just avoid using them for screenshots. In case of releases running out of sync, you're dealing with either missing sections or different frame rates. Refer to the [Advanced Tutorial TBD] on how to deal with those cases.
   2. Once all releases you want to compare are in sync, right click their tabs and add them to a single group. Make sure the Apply offsets option is checked
      AvsPmod will now remember the frame offsets between tabs and automatically seek to the right frames on a shared timeline.
5. 1. To reasonably compare screenshots, we need all frame sizes to be of the same resolution. You should have noticed any differences already while syncing up the releases.
      If all resolutions match, skip this step. If they don't, find out the highest resolution by looking at the bottom right of the status bar.
      The information will be in the following format: FrameWidth x FrameHeight (AspectRatio) - Framerate fps - Colorspace
   2. Take the largest width and height among the releases and find the following section in the scripts for all releases that need to be resized:

      target_width = w
      target_height = h

      Replace w with the largest FrameWidth and h with the largest FrameHeight
   3. Hit F5 to refresh the video preview and confirm everything works as expected.
6. Select frames for the screenshot comparison
   1. Find a frame suitable to convey significant information about an aspect of video quality.
      Pick:
      • one or two frames of a structure that represents what is on the screen most of the episode (to give a general idea about the video quality)
      • frames from dark scenes
      • high-detail frames / frames with fine structures
      • frames with low contrast
      • frames from high-motion scenes
      • a frame or two from a fade-in/fade-out
      • frames that display video artifacts in one ore more releases (refer to this thread on how to spot and identify video artifacts)
      • at least one frame from the OP and one frame from the ED
      Try to get as many B-Frames as possible. Ideally you'd want matching frame types across the different releases.
      In practice checking this gets more of a waste of time with each additional release you have to compare. Just try to keep the number of B-to-I comparisons low and make sure you don't pick only B-Frames for one release and only P-Frames for the other.
      A very easy way to avoid this issue is to not pick frames by only looking at one release. This comes with the added benefit of you being more likely to spot video artifacts in ALL releases.
      (Check the overlay on the top right of the video frame to find the frame type of the current frame.)
   2. Once you've decided on a frame, hit Ctrl+B to bookmark it.
   3. Repeat from 6a for each additional frame
      If you want to check the frames you already bookmarked, hit F2/Shift+F2 to browse through your bookmarks
7. Export screenshots
   1. Once you're done deciding on frames to screenshot (10-15 is a good number), it's time to export them to PNG.
      Go to Macros > Save image sequence (make sure Select frames is set to "Bookmarks" and Output format to "*.png")
   2. Pick an output directory and basename.
      I recommend using the basename format %06d_Groupname. This way your screenshots will (almost) always sort to comparison-friendly order by default. %06d inserts the current frame number, so make sure Add the frame number as suffix is unchecked when using it.
      

      Example of Directory+Basename: Z:\ScreenCompare\Out\%06d_Zurako

      You may also want want to check the Use always this directory option since you need to run this macro for each release seperately.
   3. Click OK. You'll get a confirmation once the export is done.
   4. Repeat from step 7a for every open tab/release

Taking screenshots with subtitles/typesetting

To come

Using the extraction script

To come

MPC-HC + madVr

To come

mpv

To come

Hosting Images

You probably want to use our image server. If for some reason that doesn't suit you, take a look at these other options:

You must host your own images. Do not hotlink images. Take a look at some of the approved uploads to get an idea of what is needed before a torrent can be approved.

• Note: Screenshots should be taken in PNG format, at the correct resolution. If the video is anamorphic, make sure screenshots are taken at display aspect ratio and resolution.
• Note 2: At least 1 shot has to show the subtitles with the style defined in the release. Click here for more detailed info.

It is recommended that you use our image hosting server BakaSHOTS as it is the most compatible with our site. Furthermore, we won't have dead screenshots when images are pruned on other hosting sites. To use BakaSHOTS just sign in using your BakaBT account.

If for some reason that doesn't suit you, PhotoBucket is an acceptable alternative.

• Note: While PhotoBucket does offer free accounts, you must use an account with a purchased subscription. The free accounts will generate a watermark over top all linked images.

Image Hosts You Shouldn't Use

• ImageShack - Resizes and degrades image quality after an unknown amount of time. Still useful for comments and temporarily uploading images before transferring them to BakaSHOTS.
• Imgur - According to their "Terms of Service" hotlinking to torrent sites is prohibited.

Adding Screenshots

Adding these tags will not do anything to your description, instead it will add screenshots just below your description using a lightbox viewer for easy viewing. If using HTML code, screenshot tags may be placed inside or outside the <html> tags, there is no difference in parsing either way.

[screenshot]screenshot code goes here[/screenshot]

Inserts screenshot thumbnails below the description. Thumbnails are resized to 150 pixels in width by default.

[screenshot=200]screenshot code goes here[/screenshot]

Inserts screenshot thumbnails 200 pixels wide below the description. Change 200 to the pixel width of your screenshot thumbnails as necessary.

You can add more than one block of screenshots, possibly with different size thumbnails.

[img]link to image[/img]

Will add one image. You can add as many as you like.

[url=link to image][img]link to thumbnail[/img][/url]

Allows you to explicitly define your own thumbnail images. If no thumbnail image is supplied it will use the linked image and resize it to whatever width is supplied for the thumbnail.

NOTE: Screenshots will not show up in the Sandbox; you have to go to the torrent description page in order to view them.

If thumbnail images are not used, the big images are loaded first, then resized to thumbnail-size on your screen. This makes image-loading really slow, and also make scrolling laggy, so as far as possible supply thumbnail images or use an image-host that auto-generates them.

BakaSHOTS Images

For details on how to insert images hosted on BakaSHOTS into descriptions, please see here: BBCode#BakaSHOTS.