Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
WorkHorse
  • 227 Replies sorted by
  • No freezing with Workhorse-Flowmotion setd, Lexar card 600x (90Mb), Autofocus single, OIS stabilizer.

  • Just one stop recording with Workh.-Nebula, Lexar 90Mb card, after four videoclips.

  • @TrackZillas @frullaccia @JayUKB

    I advanced some tasks today, and in the late afternoon I did measurements, tests and tweaking in the frame buffer and frame limit keeping same datarate for all slots.

    I think I found the GH2 G Spot! the camera went into multiple orgasms...

    120Mbps datarate in all slots, tests with the 45Mbps card in NTSC and PAL:

    30 seconds recording in SH, H, HBR, 24H and 24L without camera freeze, without stop recording. Using the 14-42mm Kit lens, auto focus disabled, OIS disabled, on the SDC4 death chart.

    Will need your tests in 80Mbps 95Mbps and 280Mbps cards to be sure about stability (I just have the 45Mbps card).

    If this thing can record some minutes in fast cards it will be the Fresh Orange Juice from the GH2!

    I will upload this patch when my tests get finished.

  • Does anyone have any screenshots they could post for a side by side comparison of this hack vs no hack? Id love to see some 100% crops of the same scene with and without the hack!

  • @joethepro we are tweaking it for more stability, so the image tests will come later. there are two videos on vimeo showing the lowlight results in previous posts from patch before these new tweaking.

  • @TrackZillas @frullaccia @JayUKB

    The patch below is the best possible 120Mbps settings for stability. It is the WorkHorse Nebula version, I chose it to test because it has big I frames, which is important to tweak the frame buffer and frame limit. It does not freeze the camera with the 45Mbps card in all modes and in ntsc or pal.

    I am in the need of tests in the 720p SH with your fast cards to know if it can record more time, in ntsc and pal (720p50 SH and 720p60 SH). Please test with OIS and autofocus disabled first.

    If it works I will have all the answers to finish all other patches with other datarates.

    Please test it with your fast cards to perceive how many time it can record in 720p60 SH and 720p50 SH and post your reports.

    Many thanks.

    120Mbps.zip
    10K
  • @apefos WOW!!! Nice man.. Will test in the morning before my video shoot. If all pans-out I will be shooting some nice slomo tomorrow. Once again thank you for all your hard work really appreciate it cheers.

  • You will be glad to know, your hack has a perfect stability in all PAL modes (1080 24H-HBR, FSH, and 720 SH-H) (OIS, AF, I, IR, AExposure, on and off). Tested indoor, nighttime, 30seconds clips, with a 90Mb Lexar card (which is not so good as a Sundisk, but quite close to it).

  • I forgot to tell you that I also tested the 80% frame rate. Everything is ok. I will do some new try in the morning.

  • Tested in the morning. No recording problems in my camera.

  • @frullaccia Many thanks for the reports. It seems I found the final tweaking, I hope. Now I will do the same things in the other datarates. Did you got stability in some minutes recording?

    @TrackZillas waiting your results!

    Guys, about PAL in NTSC lands, I live in 60Hz country, when I point the camera in PAL to the computer monitor, after press record, the image shows dark and bright strips which scroll up. Shutter 60 does not solve, the frame rate is mandatory... Under the sun there will be no problem because sun is fire, but under artificial light I think it will be no good. I will do my best in 720p60 also as I did in 720p50 in this last patch.

    This last patch is 120Mbps, highest possible in all slots. This thing working proves that the others will work also. The challenge now will be to make the 720p60 stable also in higher datarates, the camera will answer what it can do.

  • It is stable. But as my first impression, there's maybe a little more noise in this tweak, compared with the same Wkh-Neb of the last 9 september. I need to try it again. Sometimes the camera acts differently even with the same patch.

  • @frullaccia Maybe the more noise is due to the datarate is very high and then the codec preserves more the noise instead of flatten it, because higher compression tends to crush the noise.

    you need to shoot exactly the same scene, same camera adjusts and same framing, same light to perceive subtle differences.

    do you prefer more noise or less?

  • Ah ah! I prefer more detail and less noise! No, I think you are right, the difference in noise is very little, but it seems that this last patch is slightly better performing. The scene of the crime is the same: roofs, at least twenty different species of trees (a world of leaves), grass, blue sky with sunny clouds, different distances, mountains, tents, walls, chimneys, streets, flags, hanging clothes, hens: from my window you bet you can shoot real stress tests.

  • It's a good crime scene!

    I've got a feeling that boson and workhorse will merge into one... not sure yet... things are pointing that there is a sweet spot between them... if this turns into feasible, noise and detail will be balanced together.

    If this can be possible to do, I will call it the "G-SPOT" , the "G-H2O-rgasm" !

    But as I said, it is just a feeling now... I will pursue it.

  • In fact, there is some light and detail texture very appealing with Boson. While Workhorse seems a little more natural (so little differences that maybe I'm just talking silly).

  • Maybe I'm wrong, I don't know how to explain, but there is a little difference between them.

  • In tech words what happens is (in 720p):

    in workhorse I frames are bigger, this helps to record details better in good light low iso. But P frames are small, so when there is noise the P frames cannot show it perfectly. boson is the opposite, when video is in playback boson texture is more pleasing to the eye, more filmic because all frames have the noise preserved enough, but the I frames are not big enough to hold the complex details.

    more easy to perceive in low light iso 1600 or in good light in flat surfaces in shadows.

    both things together would be ideal, this is what kodak film does. in pal 720p it is more easy to do due to less frames, so more data to each frame. ntsc 720p is the challenge. the secrets are in the gop tables.

  • the g-spot (boson+workhorse togheter), details and noise in same slot, seems to be possible for 720p50 in PAL because there is enough datarate to increase both I and P frames without get into the unstable datarate territory.

    in NTSC, there is no way. no matter the gop table settings, the datarate does no increase enough to do it. in the highest possible datarate before camera freeze the amount of data to share between I and P frames are in the lower limit for both things together and chances of stability are small. one attempt to do it in ntsc is to put the boson better noise in SH slot and workhorse better detail in H slot, similar thing which boson versions already have.

  • @apefos In your testing don't forget frame sizes vary on static shots to moving shots. If you look back at many of my Cluster X settings they have been tested widely by users for all levels. Also, long ago we tested a 3 GOP outstanding setting in Spizz T6.

    with a moderate matrix but using a quantisation rate factor to define the IQ. Notice this shows Spizz T6 with QP values as low as they can get at 24p on moving water! :-)

  • Many thanks @Driftwood I saw the video, it does significant difference!

    I will do a careful look in the SpizzT6 ini file! seems to be different approach than the T7.

    much appreciated!

  • Yay!! For me, Spizz in itself is perfect, and if some legend like @apefos is still trying to improve on it, thats even better. Finding the G-spot requires some good ol' Spizz man.

  • @apefos
    Each of my settings are trials - i.e. they are all tested to work well. No single setting can be considered better than the other as each is using different variables and arguments. In the end it boils down to what kind of bitrate you want to use but of course the more applied, the more resources can be assigned to things like QP parameters and knocking out as many artefacts as you can with supreme settings. This is why Cluster X series has always strived for the best quality according to each GOP.

    Just remember to test buffers with Elecard. You don't want them to overflow or underflow / or decline rapidly which will suggest problems.

    Driftwood 'Space' and a few other settings in the past (like REDshift) strived for performance at lower bitrate settings for people making documentaries etc.

    Somewhere Ive covered all the goalposts at some time or other with my settings whether its Intra or Long GOP! But everything can be tweaked and its good to see you tirelessly keeping the GH2 settings tuned to max! Keep up the good work.

  • I am using streamparser, I did not use Elecard yet, it seems I need to learn.

    I am tweaking the buffers and frames limits testing with the camera. I consider the frame limit 10% above the highest frame in the stream when the flux is constant in the death chart, similar but not equal to 10% higher than the average i frame. this approach is working pretty amazing for image quality and stability. in general I double the frame size and type in the buffer. it works good this way even for very high datarates in the 45Mbps card. no camera freezing.

    I perceived that what makes the camera to freeze is a high i frame limit and/or a low buffer. if these two settings are correct, than no more freezing. if the card cannot handle the datarate, the camera stop recording and save the file and return to stand by mode. this means that a faster card is welcome. this is what are making my new versions more stable, including the 120Mbps last uploaded.

    at this moment I am trying to find the best balance between I and P frames in ntsc to get the best possible from 720p60, because datarate in ntsc 720 is very very tight. I already found a good setting for 720p50. I am doing this in the gop tables which distribute the datarate among the frames.

    I think this is the last thing to do from now, before upload new versions, the most stable and with best use of datarate, I hope.

    so after this I can take a look into Spizz.

    thanks!

  • @apefos Just tested your latest WorkHorse Nebula Patch all I can say is the quality is amazing those i-Frames are the highest I've seen. Using 80mbs card & manual lens 720p 50fps only got around 18 seconds, couldn't test out with my other 95mbs cards because they were formatted for the BMPCC. Images looks really awesome great work.