Topics

better dr
better low light
better gradation
better color (subjective)

so, although 3.5k has workable dual iso, this does not compensate the full sensor's advantage.

Image quality wise, 5D3 ML 3.5k and 5.7k AR 2.4 24p 14 bit lossless 14 bit color depth modes are no less than R1MX 4.5k ws RedCode 42 (42 MB/s) compression ratio 7.5 24p.
For rendering time in Davinci Resolve 18, R3D is about 3 times faster than ML raw 4k, more so than ML raw 5.7k.
Now lens is the bottle neck, instead of camera. Don't blame your camera, blame your techniques.
I suspect that once 7D2 and 5D4 have ML raw ported, they may be at least equivalent to BMPCC 6K/pro, Red Dragon 6k, or even Alexa 35 (maybe dr not so).

ML Rocks!

Using Dannes 2/16/23 version of ML 5D3 1.13

3.5K 1x1 ar 2.07 10 bit 14 bit lossless is continuous. However, pink highlight clips are about 20-30%.
UHD 1x1 ar 2.67 12 bit 14 bit lossless is continuous.
5.7k 1x3 ar 2.39 14 bit 14 bit lossless is continuous. Have not seen a single clip with pink highlight.


Using Bil's 5/25/22 version of ML 650D

UHD 1x3 AR 1.66 12 bit 14 bit lossless is continuous. Extensive field footage shows that pink highlight clips are less than 10%, and this is when the exposure is adjusted 0.5 stops lower (auto ettr in acquisition) in mlv app. Otherwise, pink highlight will not show up.

Full res 1x1 6 fps 10 bit is continuous. Pink highlight is almost on every clip.


Take home message to me is that using 14 bit modes when possible. 10 bit modes need to be well optimized and well tested before being used for field projects.


Thanks a lot to Danne and Bil and other ML gurus for their contribution to ML.

ml version: crop_rec_4k_mlv_snd_isogain_1x3_presets_2022May15.5D3113.
5D3, 50 mm stm ef lens. Canon Q menu ipb 1080p 24p. Card spanning, sd card overclocked to 160 mhz.
In full resolution live view mode, AR 2.4, 10 bit, 14 bit lossless, 7.4 fps, no fps override, the ml calculation is either green or yellow, about 100 MB/s.
Yet when I push the start button, the recording time is less than 1 second.
Turning off card spanning, and sd overclocking do not help.
Not sure what I missed here.

In anamorphic full resolution mode, AR 16:9, fps 15, 14 bit, 14 bit lossless, works perfect, manually stop after 1 minute.
If I want to change fps to 24 fps, 10 bit, 14 bit lossless, the camera freezes. I have to take out the battery. Seems the fps override does not work in this mode.

thinking to get a 70d for its dpaf.
if it is working in crop mode, maybe 3k can be continuous due to its 75-80 mb/s recording speed.
seems dpaf works in uncropped mode, but not sure if dpaf works in cropped mode.
3k 1x1 10 bit raw with dpaf for a super 16mm sensor size is sweet.
hope 1x3 modes work on 70d in the future.
ml really rocks.

I feel that I should feedback some success stories.
I tried 650D ML for about five sessions of real world events news shooting (non paid volunteer work), each session last about 2 to 4 hours, mostly from noon to afternoon, sunny day, no rain no cloud, ambient temperature kind of hot but not sweaty. 650D mode is UHD 1x3, AR 16:9, 12 bit 14 bit lossless, sd card overclocking to 240 mhz, sd card 256 gb san disk extreme pro 170 mb/s. The camera temperature was always at around 50-55 C. Later I did not pay attention to temperature anymore, yet no issue for the camera and the card.
The ml raw files were processed in mlvapp, exported to dnxhr 10 bit codec, edited in blender, finally exported to 1080p. On a 27 inch computer screen in 1080p full screen mode, image quality is so good. I would say that my little 650D ML is a baby Amira. LOL.
Thanks a lot to Bil and other ML gurus.   

just tested on 5d3, uhd 1x1, 3.5k 1x1, 14 bit /14 bit lossless modes, artificial light, iso auto 1600:
canon q menu:
ipb + small raw,
all-i + raw,
all-i + mid raw
all-i + raw/l-jpeg

the recording time is the same ball park, 50-70 frames.

i thought ipb plus raw can save out some camera memory. but seems just marginal effects.

I just tried 3x3 modes, 14bit lossless, card spanning off, sd card overclocking off (I tried both on, the same results for the 1080p 60 fps):

Canon Q menu ALL-I 1920x1080p 24p mode, 1920x1280p, 3x3, 24 fps, 55 MB/s, continuous, green
Canon Q menu ALL-I 1920x1080p 30p mode, 1920x1280p, 3x3, 30 fps, 70 MB/s, continuous, green
Canon Q menu ALL-I 1280x720p 60p mode, 1920x672p, 3x3, 60 fps, 75 MB/s, continuous, yellow
Canon Q menu ALL-I 1280x720p 60p mode, 1920x1080p 50 fps preset, 85 MB/s, 5 seconds, red, automatically off
Canon Q menu ALL-I 1280x720p 60p mode, 1920x1080p 60 fps preset, red, several seconds, automatically off

I assume that for 1920x672p, I need to destretch to get 1920x800p.
1920x1080p 60 fps at ar 2.4 supposes to work, it is 120 MB/s.

The reason that I am interested in 1080p vista vision mode, is that the lcd is realtime color and can output through hdmi to evf, and can go to 60 fps. At 24 fps or 30 fps, I can turn off the card spanning and sd card overclocking, so that there is no potential risk of overheating. The system is very robust in these modes. These modes are good for doc, news, events, corporate interview, etc.

https://www.youtube.com/watch?v=-N9dnWXYCFw
4:47
https://vimeo.com/444943732?embedded=true&source=vimeo_logo&owner=28284687
by IDA ML

I actually like the EOS M footage. It is stunning to my eyes, even in 27 inch computer screen full mode.
This is not scientific. Because Youtube may have more severe compression. But you get a feeling, since Alexa is more about hdr and color science but not on resolution.

Is there a way to tell the camera to meter on human subjects' skin tones like face and eyes as the basis, then auto ETTR and dual ISO accordingly?
I used auto ETTR and dual ISO together, and found out that they are good for the whole scene, like a street with high rising buildings under the noon sun, but the human faces may be darkened.

I just did a test comparing three different modes on 650D: crop 1920x1080 24 fps, anamorphic UHD 24 fps, crop 3K 16 fps. The full resolution mode is not working, I tried but did not test. To my big surprise, the anamorphic UHD footages on my 27 inch computer monitor in full screen mode are much sharper than those of the crop 1080p, just not at the same level. The native resolution of the anamorphic UHD is 1280x1800, just about 20-30% more than that of the crop 1080p, yet looks way more delicious.  The anamorphic UHD looks very similar to the crop 3K 16 fps, almost at the same level.

This really surprises me, because I used to think that the anamorphic modes are just for simple uprezzing, the native resolution is the real thing. Seems that the sensor records all of the information, then Canon has a binning algorithm to combine 3 horizontal pixels to 1 fat pixel. If MLV App has a transformation/debinning algorithm to reverse engineer the Canon binning process, or has an AI algorithm to guess right the original 3 pixels from this fat pixel, then the anamorphic UHD will be very close to the native UHD. This way, the binning process is more like a compression process, with the proper codec, this is actually helping to fit in the maximal writing speed of the camera. After writing in the cards, the information can be read out again by the computer software. If this is true, then binning modes are better than the crop pixel by pixel modes, because there are more information recorded by the sensor and saved in raw files.

On my 650D, I just tested that with Alex's Crop New module, in 1:1 full resolution preset, with maximal fps, 5200x1042, aspect ratio 5, I can get continuous recording with that yummy green dot. If I change the option to Canon Menu (24p) or 23.97p, the resolution changes to 5200x700, I get green dot again.

It seems that the hardware can record full resolution at 24 fps. The limiting bottle neck is the writing speed. If this is the case, adjusting aspect ratio instead of reducing fps is another approach and more practical.

On my 5D3, I only have presets of 4K 12 fps and full res 7.4 fps. I wonder if I can get full res 5.7k 24 fps with an aspect ratio of 5, like 5720x1124 24 fps. That will make 5D3 very advanced. Similarly, a preset of 4096 24p with an aspect ratio of 3 is also very useful.

Here is the idea:

5D3's stock flow is
1. full res (6k) raw video 30 p stream
2. pixel binning 3x3 to 1
3. compressing to all-i h264 codec

If we can bypass step 2 of pixel binning, we can get full res all-i h264 video clips, which should be within the CF card writing limit of 90 MBS. The computer power of the camera should be able to handle the new workflow, due to the skipping of the pixel binning step.

Cheers, ML really rocks.

I just did a quick experiment, for hdr, I chose 100/12800; then in dual iso, I chose +6ev to get only 1 stop improved dr. if I only use dual iso, I can get 3.5 stops more with +6ev.
I am thinking maybe in hdr, choose 100/1600, tehn in dual iso, I will choose +3ev.
I do mostly landscape, so movement is not a concern, but dr is. so I am thinking to get maximal dr to combine hdr and dual iso. maybe if it is successful, 5d3 can really compete alexa?
anybody tried dual iso plus hdr for raw video shooting? your input is welcome.

1. x5 zoom
2. 1920x1320 24 p raw
3. anamorphic 2x to compress 2x info, now 3840x1320
it is 4k raw now with 3:1 aspect ratio.

1) I known on 5D3 it is possible to do raw video at 1920x504 48 p, stretched in post to 1920x808 48 p. But I don't care about slow mo. I want HDR raw video.

2) Now is it possible to do raw video 48 p in HDR mode, like the guy did on 5d2 at 30p
http://www.magiclantern.fm/forum/index.php?topic=6075.0
http://vimeo.com/67501777
"Shoot RAW at 30fps, HDR video on, set ISO to taste. Extract DNGs, split in AE using time stretch by 50% (offset by a frame for opposite exposure) precompose and slow down to 50% using twixtor, export as TIFF sequence. Batch process each frame in HDR processor of choice, I prefer SNS HDR. adding a lot of patience you will obtain a merged HDR TIFF sequence."

3) Even further, can the new Dual ISO module work at the same time when raw 48 p in HDR mode is recording?

How much dynamic range with this combination achieve if possible?

ML rocks!

http://www.eoshd.com/content/10324/big-news-hands-on-with-continuous-raw-recording-on-canon-5d-mark-iii
"1:1 crop modes -

    3592 x 1320 – I get some drop frames at this res
    2880 x 1320
    2560 x 1080
    1920 x 1080 (full HD crop mode like GH2)
    1280 x 720 (the lower the resolution the more telephoto the crop)
"

According to EOSHD's article, 1:1 crop mode can do 3.5k if CF card is fast enough.
Now I am dreaming, if we have a CF card of 1000 MBps speed, is it possible for ML to intercept somewhere upstream the raw feed right after the sensor acquiring the whole 22 mb frame at 60 fps (720p allows 60 fps suggests that raw feed can do 60 fps)?

You guys rock, cheers ML!

Assuming that we have fast enough CF cards,
we can do 3.5 k raw with 5x zoom with CF card of 150 MBps (Toshiba),
can we do 6 k raw with 10x zoom?