AN "Valkyrie", GOLGOP3-13, GOP3ZILLA, natural, Freeze-less Reliable All-round GOP3 Settings :-)

bkmcwd

If I say strictly, ALL have changed. :-)

It is because change of a Deblocking tables influences all the modes. Therefore, other parameters, such as GOP tables, are readjusted. C5 is slightly made more sharp than C4.

Mirrorkisser

@bkmcwd: which c-version is best for 24p, if u say that with c5 all modes have changed? Ty

bkmcwd

@Mirrorkisser

Honestly, I am not sure, either. However, I think that comparing with "TYPE-ZERO1" is one standard since all the starting points are "TYPE-ZERO1."

Since a result changes also with a use lens, of course, an opinion may change with people. Moreover, a difference may not be recognized when using lenses other than Pany. On the other hand, I think that a difference does not come out so much in especially 24p among candidates. It may be able to be said that C5 is good in such meaning. :-)

danielvy

I tried the c5 and in HBR the camera freezed. I had to take the battery out to restore the function. The rest of the format types seems to be ok. I´ll upload in a while some footage...

bkmcwd

@danielvy

Thanks for your report! Which card did you use? And it may not freeze, if it is Candidate4.1.

danielvy

@bkmcwd i used a sdxc 64g 95mb class10... here is the test. Artifacts seen in HBR, and also freezin of the cam. The rest beautiful.

bkmcwd

Many thanks @danielvy! :-)

On what kind of scene did artifacts appear? I think it unavoidable that artifacts appears to some extent for interlace structure. About freezing, I make HBR settle down to a slight degree.

bkmcwd

I am now working on improving TYPE-ZERO2 Candidates. There are 2 points of improving element.

*The 1st is the strength of Deblocking effect.

*The 2nd is bandwith controls.

I have already mede 5 candidates for TYPE-ZERO2. As for this, I am making the variation with the 1st element.

Each feature is as follows.

C1-C2: Horizontal artifacts in HBR is reduced by change of a matrix.

C3-C5: In addition to the above, artifacts in 720p is reduced by adjustment of Deblocking tables.

From now on, I make it further stabilized by adjustment of GOP tables about the 2nd element. With this matrix, I have recognized that adjustment with the same technique as former is difficult. Since I consider "stability" to top priority, I do not give up about this.

It seems that erase artifacts in 720p cannot be finished by C5. In 720p, I am required in feedback of whether the Deblocking effect of C4.1 is enough...

Edit: In 24p, in TYPE-ZERO2 Candidates also including TYPE-ZERO1, although there are delicate differences, the difference may not almost be recognized. And I think that they are all satisfactory at stability and IQ.

danielvy

@bkmcwd I didn´t try the 4.1 setting you uploaded, i jumped from 4 to 5. But i feel that 720p in c5 is perfect, the difficulties were only in HBR mode, the setting i used was all in 0. I´ve seen that many use all in -2, maybe was a mistake from my part to leave all in 0. Waiting for your new setting, I think it´s close!

Tobsen

@bkmcwd shot with C5 today and it froze only once with the first press of the record button. After that, worked flawlessly, also in EXTele and 720p. Only had a peek so far, but the footage looks good! Once, we start editing I can provide some framegrabs.

Since Tuesday I have been shooting HBR in smooth all -2. Today first try 720p with same settings.

danielvy

@bkmcwd While we wait fot your new version, can you answer me a question... I need the best hack, meaning: Best image, highest bitrate, best color performance- but most important of all, no failures of any kind, because im going to a job and i cant fail, or miss a file, etc. I own a sdxc 64g 95mb class10. I aprecciatte yopur answer. Thanks!

bkmcwd

I am sorry that I am delayed in a reply since I have stuck to job for several of these days. :-(

@danielvy Many thanks for your feedbacks! :-)

"no failures of any kind"

IMHO, GOP3ZILLA_V3_BETA1 and GOLGOP3-13-MAX_EX_BETA7g are very stable and good quality although they are still the BETA versions.

@Tobsen Thanks for your report! :-)

Since 444 matrix of Chris differ in character from old matrices fundamentally, it is very difficult to stabilize settings except 24p. Honestly, still, C4.1 or C5 are not perfect except 24p. However, the more stable version is due to be done soon.

Lambo

Mirrorkisser

@bkmcwd: just been testing candidate 5. Very nice image. It did not span though at my sandisk extreme 45MBS in 24p with a canon fd 35mm. At a length of 4minutes 37 seconds the camera just stopped recording. Except spanning the setting works fine though, no freezes and in camera playback.

exilenorth

Nice hand shot, what am I looking at? or why?

bkmcwd

@Lambo Thanks for sample footage! Image quality looks good. :-)

@Mirrorkisser Thanks for your feedback and kind explanation in AN thread. :-)

In 24p, if 95MB/s SDXC is used, I think that span is probably possible.

Although I am now working on improving TYPE-ZERO2 Candidates, only in 24p, C1 or C5 may be the better than new one. That is because the image quality of 24p may deteriorate slightly if the Deblocking effect is optimized to 720p. Please compare after the release. I regard the following version as releasable tomorrow.

bkmcwd

Since the NTSC section was completed mostly, I am working on the PAL section now. :-)

Mirrorkisser

@bkmcwd: thanks for your Feedback. So you say i should use candidate 1 or candidate 5 for best 24p experience? Or is candidate 1 the best for 24p? I only use 24p, the other modes are more than secondary to me.

bkmcwd

@Mirrorkisser

I am not sure which for whether it is the best. It is because those to which priority should be given or which lens is used differ by each person.

In 24p, C1 and C2 are the same as TYPE-ZERO1. These are the settings made so that advice of Chris might be realized as much as possible.

C3 is not good. *Now impoving.

C4.1 is softer than TYPE-ZERO1, and C1 and C2.

C5 is between TYPE-ZERO 1 andC4.1.

If neither HBR nor 720p is related, I will use TYPE-ZERO1. This is my liking to the last. Again, since HBR and 720p of TYPE-ZERO1 were not good as it was, I made some candidates.

Mirrorkisser

@bkmcwd: i am interested in the most "filmic" or cinematic look and i use canon fd glass most of the time.

bkmcwd

@Mirrorkisser Yes, I understand your feeling well. However, conversely, I am still the position of wanting feedback . Especially about this setting, giving top priority to the concept by Chris, it has designed so that bandwidth can be used as much as possible safely. Although candidates are different also in 24p, I am not sure correctly whether it is how much different, either. I do not necessarily have so much time. Therefore, I want the feedback which compared the candidate by all means. :-)

Tobsen

@bkmcwd regarding feedback. Do you want something like Streamparser snapshots or mts files? I don't have streamparser, but access to some original mts files I put on my macbook to check right after the shoots last week.

bkmcwd

@Tobsen I appreciate your kind proposal. :-)

I am very interested in the original MTS files. However, it is your PERSONAL VIEW that I would like to hear it most. Of course, it is very important that there was a problem in what kind of situation and whether there was nothing. I would like to hear it also about whether where was good and whether where was bad. That is, I welcome all kinds of feedback. It does not trouble me too much information.

Tobsen

@bkmcwd SO far, I only looked at it briefly after the last shoot. It all looked good to me. No clearly visible artifacts bitrates mostly more than 60mbits (according to mediainfo) and overall pleasant skintones and noise pattern. My colleaque is doing the editing end of next week and than we can look more closely. What would be the best way to share some mts files with you?

bkmcwd

@Tobsen Many thanks! :-)

"What would be the best way to share some mts files with you?"

Dropbox or something like download URL or Vimeo or... Up to you.

Howdy, Stranger!

Categories

Tags in Topic

Top Posters