Woah does your site have a READY TO BURN FILE?, that is great.

I am reading the entire thread, I see you posted countless of zip for different things

did you put all of those in your ready to burn file or would I need to DL these zips

separate?

I am reading the entire thread, I see you posted countless of zip for different things

did you put all of those in your ready to burn file or would I need to DL these zips

separate?
There's one big zip with almost all of the files.

http://www.w6rz.net/allpatterns.zip

Be careful, it unzips to around 4.7 GB.

Ron

dr1394,

http://www.w6rz.net/overscancrop.zip works! It must be the chroma. At your convenience, can you redo the http://www.w6rz.net/interlace.zip with 4:2:0 chroma?

Thanks again.
Okay, here it is.

http://www.w6rz.net/interlace.zip

Because it's the same url, you may have to clear the cache on your browser to download the new file. The date of the new file is 4/6/2009 at 10:12pm.

Ron

Okay, here it is.

http://www.w6rz.net/interlace.zip

Because it's the same url, you may have to clear the cache on your browser to download the new file. The date of the new file is 4/6/2009 at 10:12pm.

Ron
Thank you so so so much. The file works great. My results are here as far as the Slingbox is concerned (not to confuse the Slingcatcher since that was just my source player):

The Slingbox does not maintain the NTSC/ATSC signal! Bogus! It drops half the fields. So my follow up question to you is:

"The simplest thing would be a pattern where all odd frames had a number 1 and all even frames have a number 2."

Did you make the number 1 the odd frames and number 2 the even frames? If so, the Slingbox only show the 1's. The odd frames. Only when I pause the video do I sometimes get 2's (50% chance). As soon I resume the video, it goes back to 1's.

If you have time, could you make the same pattern, but make it a 1280x720x60p video (progressive.zip?)? I wonder if the Slingbox HD drops half the fields when dealing with a progressive ATSC video. Thanks again.

Thank you so so so much. The file works great. My results are here as far as the Slingbox is concerned (not to confuse the Slingcatcher since that was just my source player):

The Slingbox does not maintain the NTSC/ATSC signal! Bogus! It drops half the fields. So my follow up question to you is:

"The simplest thing would be a pattern where all odd frames had a number 1 and all even frames have a number 2."

Did you make the number 1 the odd frames and number 2 the even frames? If so, the Slingbox only show the 1's. The odd frames. Only when I pause the video do I sometimes get 2's (50% chance). As soon I resume the video, it goes back to 1's.

If you have time, could you make the same pattern, but make it a 1280x720x60p video (progressive.zip?)? I wonder if the Slingbox HD drops half the fields when dealing with a progressive ATSC video. Thanks again.
The terms odd and even are a little ambiguous because there are different ways of numbering the lines. In compression, we like to use top and bottom. The top field contains the 1st line of the image. For the interlace.ts bit-stream, the "1" is in the top field and the "2" is in the bottom field. The pictures are coded top_field_first = 1, so the "1" field occurs before the "2" field in time.

Ron

If you have time, could you make the same pattern, but make it a 1280x720x60p video (progressive.zip?)? I wonder if the Slingbox HD drops half the fields when dealing with a progressive ATSC video. Thanks again.
Here's the 720p version.

http://www.w6rz.net/progressive.zip

Ron

Here's the 720p version.

http://www.w6rz.net/progressive.zip

Ron

Thank you again for your help. This is my results:

I place my source into 720/60p mode. When I play the file and stream in through my Slingbox, it picks either 1 or 2. If I pause it, it will pause at either 1 or 2 (50% chance), and it will maintain the number when I resume play.

It appears the hardware has trouble picking up the fields and still only wants to stream at 30fps so it has to dump the other field.

If I set my source to 720/60p and stream the interlaced file, it essentially behaves just like the progressive.ts file, since the source is scaling/deinterlacing the signal and causes the Slingbox, again, to choose between what field to stream and what field to ignore.

I would assume that this limitation is in their encoders. I recall that WMV has some interlacing limitations, maybe not VC-1. Slingmedia's eralier products are based on WMV. The Pro HD is based on H.264, but is backwards compatible.

Very sad. I figure that their older product lines (Slingbox Classic) will be limited to horrid deinterlacing and 30fps. I am hopeful that with your help, we have shed some light on their newer product lines (Slingbox Pro HD) and maybe someday enable better ATSC/NTSC and interlaced support. Thanks for your efforts.

The terms odd and even are a little ambiguous because there are different ways of numbering the lines. In compression, we like to use top and bottom. The top field contains the 1st line of the image. For the interlace.ts bit-stream, the "1" is in the top field and the "2" is in the bottom field. The pictures are coded top_field_first = 1, so the "1" field occurs before the "2" field in time.

Ron

Some other questions, since you seem so knowledgeable.

For the sake of my next questions, 60fps is 59.94 and 30 is 29.97.

What does the industry refer to NTSC as 30 fps when it's 60 fields per second? Decoders look at your interlaced.ts file and list it as 30fps? Is it really only 30 fps when the source is encoded as such where the broadcaster's source is 30fps and they want full resolution (see ex. 1)? I know they do this with telecine film, You get 24fps and maintain the film in full resolution.

Would it be more appropriate to list A/NTSC as 60i? I can see some shows on TV that want 30fps and want to utilize the full resolution for each frame.

Correct me if I'm wrong:

ex. 1

broadcaster wants to show a TV show at 30fps in full resolution:
1st field (top) is 1920x540 - frame 1
2nd field (bot) is 1920x540 - frame 1
frame 1 1920x1080 complete
3rd field (top) is 1920x540 - frame 2
4th field (bot) is 1920x540 - frame 2
frame 2 1920x1080 complete
repeat to get full HD resolution for 30fps

Is the above example normal? This to me is 30fps or 30p. Is this a standard? This is like telecine, but for 30fps, not 24fps.

ex. 2

broadcaster wants to show you the super bowl in 60 fps in interlaced resolution.
1st field (top) is 1920x540 - frame 1
2nd field (bot) is 1920x540 - frame 2
3rd field (top) is 1920x540 - frame 3
4th field (bot) is 1920x540 - frame 4

Is the above example normal? This to me is 60fps or 60i. Is this a standard?

Why does the industry use the identification 30i in some cases? *If* ex. 2 is identified as 30i, then why isn't 720p identified as 30p. It just seems inconsistent here.

Am I making any sense here? Thanks again.

Interlace is for you to grok in your own personal way. To me, it's 30i. That's because each field is not only different in time, but also different in spatial offset (by one line). Two fields make a frame, especially when there's no motion.

For example, the two patterns I just created for you are not exactly the same. In the interlaced pattern, the "1" is one line above the "2". In the progressive pattern, the "1" and "2" are on the same lines.

Just FYI, the creation of the two patterns was entirely different. For the interlaced pattern, I created two RGB images in Photoshop. One 1920x540 image with the "1" and one 1920x540 image with the "two". Then I converted the RGB images to YCbCr. The final step was to "weave" (one line from one image followed by one line from the other image) the two fields into one frame (which I repeated 900 times to create a 30 second clip). The final output (that gets loaded on an uncompressed file server) is 900 interlaced 1920x1080 frames.

The progressive pattern was also created from two RGB image. However, they are two full size 1280x720 images of a "1" and a "2". After conversion to YCbCr, the images are just strung together alternately. That is, a frame of "1" followed by a frame of "2". The final output (that gets loaded on an uncompressed file server) is 1800 progressive 1280x720 frames.

Ron

Interlace is for you to grok in your own personal way. To me, it's 30i. That's because each field is not only different in time, but also different in spatial offset (by one line). Two fields make a frame, especially when there's no motion.


What if there is motion between the fields for all frames? Would you still consider this as 30i? There seems like there should be a distinction between this to avoid confusion when the material needs to be deinterlaced. If there is no motion between mated fields, then the decoder should be smart enough to create a 30p presentation otherwise it should scale/deinterlace to 60p.

Just FYI, the creation of the two patterns was entirely different. For the interlaced pattern, I created two RGB images in Photoshop. One 1920x540 image with the "1" and one 1920x540 image with the "two". Then I converted the RGB images to YCbCr. The final step was to "weave" (one line from one image followed by one line from the other image) the two fields into one frame (which I repeated 900 times to create a 30 second clip). The final output (that gets loaded on an uncompressed file server) is 900 interlaced 1920x1080 frames.


Ah, I get it! This is a digital encode/decoder thing. So a 30i encode is really a 30p encode (for sake of argument), BUT when it is decoded it shows top field first, then bottom, and repeat. This ends up producing a 60 images per second on interlaced displays. On a computer or progressive TV, you are at the mercy of the decoder in how it will deinterlace the stream. Interesting. Please correct me if I'm wrong here.

My last question is:
How does one determine if an encode is 30i or 30p? It seems like computers and mediaplayers just say, 29.97 fps. I'm left to guess.

Thanks again for all the great information and responses.

Ron, I know that you said using your SD encoder is more of a pain (and that you're very busy) but any chance for SD versions of the 75% Rec601 color window patterns to play on non-HD DVRs?

cheers and thanks again for all the patterns,


--tom
Finally got around to it:

http://www.w6rz.net/sdpatchroma.zip

It has 75% color windows followed by 100%.

Ron

What unique patterns, Ron, thank you !!

Just working on a way to visually show the differences between Catalyst ControlCenter's deinterl. options using DXVA (ati's HD GPU series) *here (http://www.dvbviewer.info/forum/index.php?s=&showtopic=31789&view=findpost&p=241036)* (in German).
But on the pics you may see what I mean - the differences are very slightly noticeable.

Using your "Alternating black/white 1, 2, 3 and 4 pixel strips with moving interlaced bar 1920x1080 (http://www.w6rz.net/vertrezmotion.zip)", is a better way to show the diff. between just Vector- and MotionAdaptive:

http://img172.imageshack.us/img172/5734/allixq.png (http://imageshack.us)

See the white line near Schumachers car? (out of your "sdpaltest.zip" but same with 1080i proportional to the number of horiz. lines 576->1080)
"Normal" interlaced video shows the effect only with such diagolal bright lines moving somehow diagonal (or vertical?) and not too fast.

So I made a pic how a pattern could look like, maybe if it was...
- first interlaced still life for some seconds (to show Bob + Adaptive)
- then moving (like your "Film vertical and horizontal resolution test") but - diagonally left down to right up and back - or up-down (to show Wave, Motion + VectorAdaptive)
...this should finally blow up the ati-mashine?? :-)

cut out, resized:
http://img229.imageshack.us/img229/9148/small.png (http://imageshack.us)

Moving interlaced bar? Don't know, maybe the moving pic is enough..
--- Europe is PAL country so 25 fps interlaced would be a good thing !

If that is just too much I like your work anyways.. perhaps just a 25fps version of "Alternating black/white 1, 2, 3 and 4 pixel strips with moving interlaced bar 1920x1080"?

Craig

Finally got around to it:

http://www.w6rz.net/sdpatchroma.zip

It has 75% color windows followed by 100%.

Ron

Thanks so much Ron. It worked perfectly on the SD TIVO.

Once again, thanks for all the effort on these patterns.


--tom

Just to start the ball rolling, here's an extended gamut color bars pattern.

http://www.w6rz.net/bars709_110.zip

The color bars are split into 110% (except for the white bar at 100% and the yellow bar at 107% due to having to limit the Y value to 235) on top, 100% in the middle and 90% on the bottom.

Here's the pattern in 16 to 235 RGB (just to show the layout).

http://www.w6rz.net/barsxv.png

Here's the same pattern in 0 to 255 RGB.

http://www.w6rz.net/barsx.png

Ron

..did it myself now, Patterns 1080i + SD but thanks Ron, yours helped to give ideas! Any suggestions how to make it better?
Here's the thread (http://www.avsforum.com/avs-vb/showthread.php?t=1157287)

craig

Hi,

i wonder if there is any screenshot or description what vertrezmotion.ts should look like with different deinterlacers, so how can I compare it to what I actually get?