This is a color grading test with Sanity 4.1 by Ralph B - PAL version @ HBR 25p settings. Password is "123qwe!"

Sorry, wrong post. Posting in the right place!

@bkmcwd @towi

Looking at the 720p GOP tables we've tested so far I think we shoud stay within this pattern:

Opt1: 1, X, 0, 0, Y, 0

Opt2: 1, Y, 0, X, 0, 0

Yes!!! And 1+X+Y=the multiple of 3

In recent GOLGOP3 patch files, I see the following GOP patches:

720p60 GOP Size=6

720p60 Opt1 GOP Table=1, 19, 0, 0, 43, 0

720p60 Opt2 GOP Table=1, 43, 0, 19, 0, 0

I'm curious how these numbers were determined, as the sum doesn't add up to a multiple of 720p mode's 6-frame GOP?

In practice, I do see an effect on the 720p encoder using these GOP Tables, however, it appears to be dysfunctional. Although I-frame sizes in SH mode can range up to 500KB, P and B-frames sizes are depressed, and the QP values are too high, ranging from 20-25 even when setting the 720p Quantizer to 18.

Do you mean 'sum' when you say 'checksum'? Checksums are a specific thing.

@LPowell

"I take it these GOP Tables were found by experimentation and testing"

yes, exactly! Very, very time consuming ... but I think it was worth the trouble...

@LPowell

"I'm curious how these numbers were determined, as the sum doesn't add up to a multiple of 720p mode's 6-frame GOP?"

I am changing the GOP table not only in 720p but in all the modes. This is because it discovered by chance that the bit rate which actually appears was controllable when the table was changed. Originally, on a scene stressful in 3GOP, as for 24p of Q13, frame size will become small. However, it turned out that this is avoidable by changing GOP table. On the extension, various combination has been tried in other modes. It is the result of exploring together with towi about the table of 720p. It was made value where a buffer is stabilized as his explanation.

If surely the multiple of 6 is used, I will also imagine whether it is the better. However, since frame size became large when there was as much as possible little sum total value, priority was given there and the multiple of 6 was given up.

As stated from the beginning, I do not understand about whether this method is right. However, since it is satisfactory while using, I have adopted now.

@bkmcwd @towi Your are making a lot of cool stuff, can´t follow you trough all your settings. Has GOLGOP3-13_SednaAQ1C_V2 also the latest optimized settings for 50p? Or the NTCS and PAL settings interfere each other?

@Kihlian

Thanks! :-)

"Or the NTCS and PAL settings interfere each other?"

I think that I arranged so that both could be used. However, testing alone supposing all the cases is impossible. Please try and tell me comment.

@jard667

Thanks for feedback! :-)

I have not done the strict test with your card. However, I surmise that all of the GOLGOP13-13 series cannot be spanning in 24p with the card. In order that, as for this SednaAQ1C version, my friend might use 24H, I made. Although he is also using the same card as you, he does not care about whether span is possible.

Tested Cinema 24p for spanning. 14-140 pany, F4, ISO3200, SS25, 32GB Transend class 10. Hung it from a pipe in the low light stairwell and made it swing around while recording brick walls, wood beams and such (lots of texture and shadows). 5:02 later and 3.99GB it quit recording. Video was fine (no write errors, in camera playback) but no span.

These settings work fine for all modes. bkmcwd and me were working together on these, so they are optimized for NTSC as well as for PAL modes: http://personal-view.com/talks/discussion/comment/54523#Comment_54523

Me personally I've added theses values to all 3 patches:

• 1080p24 GOPx2=6

• 1080i60 GOPx2=12

• 1080i50 GOPx2=12

• 720p60 GOPx2=12

• 720p50 GOPx2=12

I believe they improve spanning. I am not sure, though.

BTW: the latest GOLGOP13-SednaC doesn't work for me in HBR/25p mode.

@PixCanFly

you've picked exactly the one setting that should be ignored... use one of these: http://personal-view.com/talks/discussion/comment/54523#Comment_54523

Maybe I'll come up with a "motion optimized" 25p GOP3ZILLA sooner or later. But GOLGOP13 is great with motion in HBR/25p anyway... so I don't think a modfied version of GOP3ZILLA is usefull...

Yesterday I've started to test GOP3ZILLA with P-Frames instead of B-Frames in HBR/25p. Looks very, very promising... but it's not yet stable (and unfortunately it's not easy to make it stable ...)

@Bonzai

Thanks for feedback, Bob! :-)

About span, I am not taking this setting into consideration at all. Although span can probably be made possible, I think that it will become impossible to harness the character of SednaAQ1C.

@soulkeeper

4.29GB is correct. See for instance here: http://www.personal-view.com/talks/discussion/comment/54351#Comment_54351

@towi good to know it's ok!

Another "problem"... all of the short test clips i made with many patches(sanity 4.1,cake 2.0,mysteron,sedna AAQ1) in hbr/25p ,i played them flawlessly with PLEX on mac... i tried to play now using GOLGOP13 4.13 and there were jaggies all over to everything moving...i had to select deinterlace in plex options... Why does this happens only with this patch?