Got some results! Overclocked sd-card bus from 72 MHz to 84 MHz.
But I have one problem: I don't know, how it affects read/write speed (no tools to check real I/O speed).
[UPDATE] To check I/O real speed I use recording in MJPEG (I change jpeg quality - it gives me determined bitrate). Just tried to change sd-card bus frequency to 10 MHz - it really works, "slow card" message appears immediately => my script with changing bus frequency works correctly. Now we just need to understand how high can we go.
Hey, have some news. H.264 is not limited by sd-card speed. I know, these words seems funny, but it's real. When I make card slower, H.264 recording says "slow card". But when I increase bitrate instead of slowing sd-card, I get camera frozen after some seconds of recording.
What can be the reason of this behavior?
I think, that there is a buffer between DSP and File-writing threads, that becomes full before file-writing thread catches blocks. I tried to change some bytes to non-instruction to watch behavior of camera when it reaches it. Same result. Camera is just frozen.
Yep, you have lot of fun still awaiting you :-)
Finally bought a used E1, and it seems to have a native 4:3 mode in 2k, I wonder if the resolution can be bumped up like the GH4 all photo mode
@storyboardcreativity, thank you very much for this scripts! I've founded that HD resolutions don't freeze with CBR. Bitrate is lower (65543kbps), but it is far better than it was (15Mbps). And now HD resolutions are useable. And I lowered VBR multiplier for 4K resolutions to match my Samsung EVO cards. It is interesting, because this cards can handle 135Mbps from Yi4K+ but this camera says "slow card" Anyway it is way better than it was before. And can you help me to find 4K DCI @25fps mode?
Is it possible to set chroma 422 in mjpeg mode? Perhaps 420 error is only for h.264
@zcream, I worked on that for some time, but I need to disassemble Ambarella's DSP microcode in firmware. I can't understand, what is the instruction set in it.
@Leemuzhko, I now don't have much time for working on this firmware, sorry.
Additional info: I have a little inside from Z Cam - I spoke to one of the developers and he said that they don't like Ambarella SoC's very much. They said that their SDK is very buggy and they don't get any support from Ambarella. Now Z Cam change their vendor because of these problems.
I have a little inside from Z Cam - I spoke to one of the developers and he said that they don't like Ambarella SoC's very much. They said that their SDK is very buggy and they don't get any support from Ambarella. Now Z Cam change their vendor because of these problems.
You know that Z-CAM had been founded by ex Ambarella engineers, right? :-)
The chipset it's for Consumer Cams, Codec embedded, the E1 Sensor have 10bit ,the Chipset can only 8 Bit, For PRO Res codec Need modern Chipset look of teh new FF E2 8K pro res .....4x Bit Rate from 4k ....16x Bit Rate from 1080P...... With e1 100k1gop Hack my e1 Record 0n 106mbps on 8bit Codec....... Pro Codec Need pro Hardware !!!!!!
@MikeLinn, that's a very strange situation lol
hi people @storyboardcreativity the monitor of the camera does not allow to see the focus comfortably, is it possible to add focus peak fuction as the image shows attach?
on the other hand photo in 16: 9 format please
Any plans to improve the quality photos? focus improvements... any others
his contributions are great
I don’t think peaking is possible as it probably require coding new features. I’m just hoping for an open gate mode that captures full width of the sensor.
Thank you for your answers
The difficulty is not having an external monitor for continuous video recording. I have noticed how deficient the automatic focus mode, I hope we can do something
in relation to having DNG photographs in 16: 9 format do you have any ideas?
any improvement for the great compression and lack of detail in dng?
@zjavier, I'm working on video side now - have too awful quality in zlog mode (not because of 8-bit mode, but because of too low h264 compression quality). Footages after grading become garbage.
thank you very much @storyboardcreativity, maybe out zlog, only normal mode you can make something, let's have faith of all the good things that come
Just got all MN34230 sensor input modes in E1:
[0]: Sensor Offset: 0 x 0 Rect Size: 4644 x 3506 H: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) V: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) SummingFactor = 1
[1]: Sensor Offset: 0 x 400 Rect Size: 4640 x 2704 H: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/2) V: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/2) SummingFactor = 1
[2]: Sensor Offset: 0 x 0 Rect Size: 4638 x 3504 H: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/3) V: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/3) SummingFactor = 1
[3]: Sensor Offset: 0 x 18 Rect Size: 4638 x 3470 H: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/3) V: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/5) SummingFactor = 1
[4]: Sensor Offset: 0 x 2 Rect Size: 4640 x 3500 H: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/2) V: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/2) SummingFactor = 1
[5]: Sensor Offset: 0 x 24 Rect Size: 4638 x 3456 H: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/3) V: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/9) SummingFactor = 1
[6]: Sensor Offset: 0 x 72 Rect Size: 4638 x 3360 H: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/3) V: AMBA_SENSOR_SUBSAMPLE_TYPE_BINNING (1/20) SummingFactor = 1
[7]: Sensor Offset: 0 x 528 Rect Size: 4644 x 2450 H: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) V: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) SummingFactor = 1
[8]: Sensor Offset: 0 x 446 Rect Size: 4644 x 2614 H: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) V: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) SummingFactor = 1
[9]: Sensor Offset: 0 x 672 Rect Size: 4644 x 2160 H: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) V: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) SummingFactor = 1
[10]: Sensor Offset: 0 x 0 Rect Size: 4644 x 3506 H: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) V: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) SummingFactor = 1
[11]: Sensor Offset: 0 x 528 Rect Size: 4644 x 2450 H: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) V: AMBA_SENSOR_SUBSAMPLE_TYPE_NORMAL (1/1) SummingFactor = 1
all the photos whatever they are in dng have a size of 30.8 mb ... this seems strange to me ... I think something is not optimized ... the truth the quality should be better in less file size
@storyboardcreativity so it can capture 4:3 with 4644x3506, what would be the max frame rate?
@pureaxis, according to MN34230 datasheet: "Full scan 10bit mode (22.5 fps)"
hmm I'd image frame rate would go up if its 8bit or pixel binning
Checked zLog mode with MJPEG - much better (with higher quality percent is much-much better)
Just added patch to repository with ability to force MJPEG quality percent.
To use you have to use "Force MJPEG patch" with commands from "Force MJPEG quality patch" (you enable MJPEG and change quality). If you use quality percent patch without forcing MJPEG, nothing will happen.
WARNING: If you set too high MJPEG quality percent camera will stop recording with "slow card" message (100% quality can cause ~400-500 Mbps bitrate - impossible for sd-card).
With the hack one should consider a few things the camera gets a lot hotter with the 100mbs1gop hack. Attention Read the data sheet of the sensor !! The Hotter the sensor the more noise The Themeratur is significantly higher ! I'm getting an IR Digital Thermometer tonight and will make a video about it tomorrow .. I will also remove the battery from the camera and measure the power consumption without and with hack.
I'm at a small rock band at the mixer and dokomentation with the E1 our appearances ,while the E1 is up to 5 hours running on the power supply with Accu. Last Night Record 6x64Gb sd 2x128Gb Sd !! full
On this Time Record Without Hack ... So far with hack the risk is too high for me on this Time .
There are some E1 the heat dead Died weak point internal power supply and sensor is at 4K30p like times over 50 degrees C Hot.
I test bevor i kill my cam --- it by quickly removing the lens and IR measurement
Thanks Storyboard for fantastic work ....all other ...You Know Hacking it´s Yor Risc
It looks like you're new here. If you want to get involved, click one of these buttons!