Does anyone have a 5D and a UDMA card to test an exciting new development? The attached build has a setting for qscale to control the bit rate: the default is -8, which is ~45 Mbps. My non-UDMA Sandisk Extreme III card can't handle -12 == 64 Mbps for more than a few seconds before the buffer fills up. (It can, however, handle qscale of +8 == 7 Mbps just fine... I see a trip to Adorama in my near future.)
To adjust the bit rate, go to the "Debug" menu and click on the qscale option. It will count down to -16 and then wrap to +16 so you can try different settings. Note that this also switches the camera into a fixed bit-rate setting instead of the default VBR.
On Sun, Nov 28, 2010 at 3:14 PM, Sean Simpson <simpson...@gmail.com> wrote: > works on my SanDisk UDMA 60mb/s up to -14 and -15 where it is barely > tolerable. -16 is unusable as it breaches the 60mb/s max of my card.
> -Sean
> On Sun, Nov 28, 2010 at 2:27 PM, Trammell Hudson <hud...@osresearch.net>wrote:
>> Does anyone have a 5D and a UDMA card to test an exciting new >> development? The attached build has a setting for qscale to control >> the bit rate: the default is -8, which is ~45 Mbps. My non-UDMA >> Sandisk Extreme III card can't handle -12 == 64 Mbps for more than a >> few seconds before the buffer fills up. (It can, however, handle >> qscale of +8 == 7 Mbps just fine... I see a trip to Adorama in my >> near future.)
>> To adjust the bit rate, go to the "Debug" menu and click on the >> qscale option. It will count down to -16 and then wrap to +16 so >> you can try different settings. Note that this also switches the >> camera into a fixed bit-rate setting instead of the default VBR.
>> To post to this group, send email to ml-devel@googlegroups.com >> To unsubscribe from this group, send email to >> ml-devel+unsubscribe@googlegroups.com<ml-devel%2Bunsubscribe@googlegroups.c om> >> For more options, visit this group at >> http://groups.google.com/group/ml-devel?hl=en
On Sun, Nov 28, 2010 at 03:15:48PM -0500, Sean Simpson wrote: > works on my SanDisk UDMA 60mb/s up to -14 and -15 where it is barely > tolerable. -16 is unusable as it breaches the 60mb/s max of my card.
The cards are rated to 60 MB/s, while the video is only 60 Mb/s. That factor of 8 should be plenty of overhead, although I have seen many reports that the cards have far less bandwidth than "rated".
I have just purchased a Lexar 16 GB UDMA 400x card that is supposed to be good for 60 MB/s as well. My results have been inconsistent as to the maximum quality value it will support; sometimes -15 works consistently, but othertimes -13 fails.
Are you shooting 24, 25 or 30p? I did a quick test with -8 through -16 and posted the results on the wiki:
> On Sun, Nov 28, 2010 at 03:15:48PM -0500, Sean Simpson wrote: > > works on my SanDisk UDMA 60mb/s up to -14 and -15 where it is barely > > tolerable. -16 is unusable as it breaches the 60mb/s max of my card.
> The cards are rated to 60 MB/s, while the video is only 60 Mb/s. > That factor of 8 should be plenty of overhead, although I have seen > many reports that the cards have far less bandwidth than "rated".
> I have just purchased a Lexar 16 GB UDMA 400x card that is supposed > to be good for 60 MB/s as well. My results have been inconsistent > as to the maximum quality value it will support; sometimes -15 > works consistently, but othertimes -13 fails.
> Are you shooting 24, 25 or 30p? I did a quick test with -8 through > -16 and posted the results on the wiki:
> To post to this group, send email to ml-devel@googlegroups.com > To unsubscribe from this group, send email to > ml-devel+unsubscribe@googlegroups.com<ml-devel%2Bunsubscribe@googlegroups.c om> > For more options, visit this group at > http://groups.google.com/group/ml-devel?hl=en
I am getting buffer bar at only at one block especially when I move the camera from left to right, back and forth. I was able to record for 2mins+ on -12 on qscale, the card is sandisk extreme III 30MB/s type, 4GB.
On Sun, Nov 28, 2010 at 11:27 AM, Trammell Hudson <hud...@osresearch.net>wrote:
> Does anyone have a 5D and a UDMA card to test an exciting new > development? The attached build has a setting for qscale to control > the bit rate: the default is -8, which is ~45 Mbps. My non-UDMA > Sandisk Extreme III card can't handle -12 == 64 Mbps for more than a > few seconds before the buffer fills up. (It can, however, handle > qscale of +8 == 7 Mbps just fine... I see a trip to Adorama in my > near future.)
> To adjust the bit rate, go to the "Debug" menu and click on the > qscale option. It will count down to -16 and then wrap to +16 so > you can try different settings. Note that this also switches the > camera into a fixed bit-rate setting instead of the default VBR.
> To post to this group, send email to ml-devel@googlegroups.com > To unsubscribe from this group, send email to > ml-devel+unsubscribe@googlegroups.com<ml-devel%2Bunsubscribe@googlegroups.c om> > For more options, visit this group at > http://groups.google.com/group/ml-devel?hl=en
On Sun, Nov 28, 2010 at 9:14 PM, Gilbert M <dcr...@gmail.com> wrote: > I am getting buffer bar at only at one block especially when I move the > camera from left to right, back and forth. I was able to record for 2mins+ > on -12 on qscale, the card is sandisk extreme III 30MB/s type, 4GB.
> On Sun, Nov 28, 2010 at 11:27 AM, Trammell Hudson <hud...@osresearch.net>wrote:
>> Does anyone have a 5D and a UDMA card to test an exciting new >> development? The attached build has a setting for qscale to control >> the bit rate: the default is -8, which is ~45 Mbps. My non-UDMA >> Sandisk Extreme III card can't handle -12 == 64 Mbps for more than a >> few seconds before the buffer fills up. (It can, however, handle >> qscale of +8 == 7 Mbps just fine... I see a trip to Adorama in my >> near future.)
>> To adjust the bit rate, go to the "Debug" menu and click on the >> qscale option. It will count down to -16 and then wrap to +16 so >> you can try different settings. Note that this also switches the >> camera into a fixed bit-rate setting instead of the default VBR.
>> To post to this group, send email to ml-devel@googlegroups.com >> To unsubscribe from this group, send email to >> ml-devel+unsubscribe@googlegroups.com<ml-devel%2Bunsubscribe@googlegroups.c om> >> For more options, visit this group at >> http://groups.google.com/group/ml-devel?hl=en
The bit-rate is always a little variable with h264, try recording with
the lenscap on and see how long you can record.
If there is less information to record the bitrate on the file will go
down. with a fast moving shot it will go up.
On Nov 28, 10:26 pm, Trammell Hudson <hud...@osresearch.net> wrote:
> On Sun, Nov 28, 2010 at 03:15:48PM -0500, Sean Simpson wrote:
> > works on my SanDisk UDMA 60mb/s up to -14 and -15 where it is barely
> > tolerable. -16 is unusable as it breaches the 60mb/s max of my card.
> The cards are rated to 60 MB/s, while the video is only 60 Mb/s.
> That factor of 8 should be plenty of overhead, although I have seen
> many reports that the cards have far less bandwidth than "rated".
> I have just purchased a Lexar 16 GB UDMA 400x card that is supposed
> to be good for 60 MB/s as well. My results have been inconsistent
> as to the maximum quality value it will support; sometimes -15
> works consistently, but othertimes -13 fails.
> Are you shooting 24, 25 or 30p? I did a quick test with -8 through
> -16 and posted the results on the wiki:
> The bit-rate is always a little variable with h264, try recording with > the lenscap on and see how long you can record. > If there is less information to record the bitrate on the file will go > down. with a fast moving shot it will go up.
> On Nov 28, 10:26 pm, Trammell Hudson <hud...@osresearch.net> wrote: >> On Sun, Nov 28, 2010 at 03:15:48PM -0500, Sean Simpson wrote: >> > works on my SanDisk UDMA 60mb/s up to -14 and -15 where it is barely >> > tolerable. -16 is unusable as it breaches the 60mb/s max of my card.
>> The cards are rated to 60 MB/s, while the video is only 60 Mb/s. >> That factor of 8 should be plenty of overhead, although I have seen >> many reports that the cards have far less bandwidth than "rated".
>> I have just purchased a Lexar 16 GB UDMA 400x card that is supposed >> to be good for 60 MB/s as well. My results have been inconsistent >> as to the maximum quality value it will support; sometimes -15 >> works consistently, but othertimes -13 fails.
>> Are you shooting 24, 25 or 30p? I did a quick test with -8 through >> -16 and posted the results on the wiki:
On Mon, Nov 29, 2010 at 11:56, Gilbert M <dcr...@gmail.com> wrote: > Are my emails going through? I am not receiving anything.
> On Nov 29, 2010 3:32 AM, "Lars Steenhoff" <larssteenh...@gmail.com> wrote: >> The bit-rate is always a little variable with h264, try recording with >> the lenscap on and see how long you can record. >> If there is less information to record the bitrate on the file will go >> down. with a fast moving shot it will go up.
>> On Nov 28, 10:26 pm, Trammell Hudson <hud...@osresearch.net> wrote: >>> On Sun, Nov 28, 2010 at 03:15:48PM -0500, Sean Simpson wrote: >>> > works on my SanDisk UDMA 60mb/s up to -14 and -15 where it is barely >>> > tolerable. -16 is unusable as it breaches the 60mb/s max of my card.
>>> The cards are rated to 60 MB/s, while the video is only 60 Mb/s. >>> That factor of 8 should be plenty of overhead, although I have seen >>> many reports that the cards have far less bandwidth than "rated".
>>> I have just purchased a Lexar 16 GB UDMA 400x card that is supposed >>> to be good for 60 MB/s as well. My results have been inconsistent >>> as to the maximum quality value it will support; sometimes -15 >>> works consistently, but othertimes -13 fails.
>>> Are you shooting 24, 25 or 30p? I did a quick test with -8 through >>> -16 and posted the results on the wiki:
>> To post to this group, send email to ml-devel@googlegroups.com >> To unsubscribe from this group, send email to >> ml-devel+unsubscribe@googlegroups.com >> For more options, visit this group at >> http://groups.google.com/group/ml-devel?hl=en
> To post to this group, send email to ml-devel@googlegroups.com > To unsubscribe from this group, send email to > ml-devel+unsubscribe@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/ml-devel?hl=en
I tried a bit different experiment tonight. Put camera on slider. I
shot three different clips, on with standard firmware, one with magic
lantern at -8 and one with ML at -12. I slid across a background in
focus of various printed cloth from my wife's sewing scraps. Now I
didn't get real strict as to exact timing and speed, but this is my
result. Results
Standard Firmware 42 sec ---> 237,326 5650
KB per second.
ML Set at -8 44 sec ---> 247,534 5625
kb per second.
ML Set at -12 36 sec ---> 153,011 4250
kb per second.
Seems like my test are reducing bit rate ?
By the way -15 crashed my 133x Kingston Elite Pro card
On Nov 29, 3:31 am, Lars Steenhoff <larssteenh...@gmail.com> wrote:
> The bit-rate is always a little variable with h264, try recording with
> the lenscap on and see how long you can record.
> If there is less information to record the bitrate on the file will go
> down. with a fast moving shot it will go up.
> On Nov 28, 10:26 pm, Trammell Hudson <hud...@osresearch.net> wrote:
> > On Sun, Nov 28, 2010 at 03:15:48PM -0500, Sean Simpson wrote:
> > > works on my SanDisk UDMA 60mb/s up to -14 and -15 where it is barely
> > > tolerable. -16 is unusable as it breaches the 60mb/s max of my card.
> > The cards are rated to 60 MB/s, while the video is only 60 Mb/s.
> > That factor of 8 should be plenty of overhead, although I have seen
> > many reports that the cards have far less bandwidth than "rated".
> > I have just purchased a Lexar 16 GB UDMA 400x card that is supposed
> > to be good for 60 MB/s as well. My results have been inconsistent
> > as to the maximum quality value it will support; sometimes -15
> > works consistently, but othertimes -13 fails.
> > Are you shooting 24, 25 or 30p? I did a quick test with -8 through
> > -16 and posted the results on the wiki:
Having finished shooting for Christmas, I updated to 0.1.9 with the variable bit rates and have been trying it out this morning. Using a Sandisk Extreme III 320 MB/s card I have only once had a cut out after a few seconds at -16. On all other occasions at rates of -12 to -16 I achieved 30 seconds plus recording with no sign of buffer overflow. On most of these recordings I used whip pans.
> To post to this group, send email to ml-devel@googlegroups.com > To unsubscribe from this group, send email to > ml-devel+unsubscribe@googlegroups.com<ml-devel%2Bunsubscribe@googlegroups.c om> > For more options, visit this group at > http://groups.google.com/group/ml-devel?hl=en
> Having finished shooting for Christmas, I updated to 0.1.9 with the > variable bit rates and have been trying it out this morning. Using a > Sandisk Extreme III 320 MB/s card I have only once had a cut out after a few > seconds at -16. On all other occasions at rates of -12 to -16 I achieved 30 > seconds plus recording with no sign of buffer overflow. On most of these > recordings I used whip pans.
>> To post to this group, send email to ml-devel@googlegroups.com >> To unsubscribe from this group, send email to >> ml-devel+unsubscribe@googlegroups.com<ml-devel%2Bunsubscribe@googlegroups.c om> >> For more options, visit this group at >> http://groups.google.com/group/ml-devel?hl=en
Here are the the results of my testings with the different qscale
settings.
Because the embedded vimeo clips are recompressed it will be harder to
see the difference, if you download the source files and apply a grade
the differences will become noticable.
All original files can be downloaded at the vimeo links.
