Messages

Strange one, those links don't work for me in Chrome when clicking direct from the forum links, but work in Firefox.  Also if I go to https://pel.hu/down/ I can click and download in Chrome as normal, so there is some problem in Chrome or at my end.

Anyway, back on 1.1.3 and all seems good, thanks

I finally found it here:

https://www.touslesdrivers.com/index.php?v_page=23&v_code=34540

Really dodgy looking site though

Top of page -> Downloads -> Download nightly builds -> 5D Mark III 1.2.3 -> Installation and or -> Useful links.

Those links from the forum are not working for me, neither version downloads:

http://pel.hu/down/eos5d3-v123-win.zip
http://pel.hu/down/eos5d3-v113-win.zip

https://bitbucket.org/Dannephoto/magic-lantern_dannephoto_git/downloads/

It appears that that repository of Danne only contains ML firmwares, not Canon ones.

So, if anyone else could suggest where to grab that 1.1.3 file it would be a great help

EDIT: looks like those forum links point to the CHDK repository of Canon firmwares, and it seems like pel.hu is down for now

Wouldn't the Quality in RAW be the only file type you can edit in mlv app?

Yes the HDMI out would be recorded in whatever format you specify when taking from the capture card.

Why don't you test both HDMI options for yourself?
And AFAIK it is 4:2:0 vs 4:2:2.
See https://www.atomos.com/compatible-cameras/eos-5d-mark-iii and they must be talking about 1.2.x (and higher) because 1.1.x (out-of-the-box, without ML) doesn't have clean HDMI.

Yes you are right, the problem was my incorrect understanding that only 4:4:4 is uncompressed.  I thought because 4:2:2 has half the chroma data it was compressed, which proves a little knowledge in the wrong hands can be dangerous. 

https://learnmediatech.com/chroma-subsampling-444-vs-422-vs-420/

On further reading it seems this change in output is of no consequence to me, therefore not wishing to compromise on the things 1.1.3 does better, it's unlikely I'll try 1.2.3 as there will be no reason

Brilliant, thank you so much, I'll get better at knowing where to look.

So if anyone reads this and is in my position, you may conclude like me that the uncompressed HDMI out in 1.2.3 is not worth throwing away all the 'display options' as mentioned above.

I like to work with a hoodeye outdoors on manual glass so digic peaking and zebras are a must.

If I understand correctly, 1.2.3 can do 4:4:4 over HDMI while 1.1.3 is going to be 4:2:2.  I'm curious to know what practical differences I should expect based on what others have experienced in the image q between the 2 over HDMI, and it could be other factors than the sampling compression

Thanks for the suggestion.  I still haven't been able to find a definitive comparison of 1.1.3 and 1.2.3 and how to choose based on your needs, so apologies if I've missed this.

I understood that from 1.2.1 onwards uncompressed HDMI output was supported.  I must confess after an hour of searching, I'm still not clear what these display settings are about, which is almost certainly me being thick, but I tried.

I would like to use the camera for raw in the field, but in the studio will be looking to run HDMI from it quite a bit

Great to hear the further feedback, and of course special thanks to Danne and everyone who helped make this miracle possible.

My 5d3 adventure got off to a good start, the thing is almost flawless and has that low shutter count.  Couldn't be happier.

It has 1.3.6 firmware, as expected, so I'm looking to get it back to 1.2.3 and get going with ML now. 

However none of the links/pages I have found actually link to the 1.2.3 .fir file, so I'm stuck for now.

Does anyone have a link?

Thanks

Dived in!  £629, 45k shutter actuations, looks in decent shape, hope the sensor's OK.  Wish me luck.

Now I have to try to forget about it for a week while I wait.  Wish me luck.  I've watched a couple more '4k' raw efforts on youtube, and it just has me marvelling at the rendering in places:

https://www.youtube.com/watch?v=FSIDg_ih5xo

https://www.youtube.com/watch?v=s81BQpbFOxs

Huge thanks again to everyone who made this possible.  It really makes a difference that the toolset ML gives you is a labour of love and is done to freely grant access to learnings and results like this.  Going out and dropping a few k's on a pro camera and getting these results wouldn't feel the same to me.

Well thanks for the reply it's appreciated.   Also I am just a hacker and el cheapo guy who gets a kick out of anything 'custom firmware'.  One reason for this is I don't like waste.  It's great if you can add pro features in software to entry level prosumer gear, because then the hacker can learn techniques associcated with these features and with a bit of effort produce results that can come pretty close to acceptable for a pro.

I don't know what you mean by "image quality ceiling"

I mean the best image quality available today, but I'm referring to the prosumer price bracket.

but it is quite obvious that you cannot compare the sensor of a 8-year old camera with the most recent sensors.

Yeah that's what I mean when I talk about the gap

I've just spent a while reviewing all the cameras that actually do 4k30 raw, and although I imagine many could be had for a similar price s/h to the 5d3, I'm highly doubtful any will come close in terms of features to explore and learn from that 5d3+ML affords.  Not to mention that the 5d3 is still a bit of a rolls royce for stills.

I guess the answer will be get a 5D3 and then add a s/h latest gen mirrorless camera in a similar price bracket to widen out options

I'm just really happy after all these years of not having time (kids, work!) that the community is still at it, and A1ex is still active

Ah thought so, but I really appreciate the replies and the vid.  After just a little of my own research, I realise my question was a bit lazy, so thanks again.

Low light performance apart, I wonder if there was a sense of how much ML Raw 14 bit closes the gap on the latest generation of sensors and processsors in terms of image quality ceiling. 

That video is a damn good advert.  What's most impressive is how the busker makes a guitar sound like a piano.

Hi,

Looking for help choosing which model to go for:

- I like the power of ML.  You could even say I'm partly doing this just to play with ML, I know it's not very pro.
- Low light performance is a big one
- Would like to get the best bang for buck for raw video
- Good stills shooting performance
- I have quiet a few m42 primes.  It's my Takumar collection I'm most concerned about, but this requirement could go if all the others are met
- Sensor crop for zooming video would be good (t2i didn't have that last time I played)
- Good resolution output for capturing (t2i / 550d I had before was not full HD)
- Willing to spend £500-700, but if something is £1k and just way better, I'm interested

Any suggestions/tips most welcome.

Thanks

I've been looking at s/h 5d mkIII's on ebay and they go for about £500 body only.

I....

http://www.komputerbay.com/terms-and-conditions 

That is at odds with the way the cards are presented on Amazon:

- COMPATIBLE WITH DSLR, SLR, AND OTHER HIGH END CAMERAS
- Comes with free plastic case compact flash and Lifetime Warranty

And:

Ultra Reliability

Backed by a lifetime limited guarantee, you can have confidence in the quality of your card and its ability to withstand some of life's toughest conditions to continue to provide you with a long-lasting, reliable memory storage solution throughout its own lifetime and that of your Compact Flash compatible device.

It even says 'professional' on the card.

Bit misleading? The text is in direct conflict with the page you linked to.

Those cards are factory seconds.  Which means a testing machine found potential problems.  When you buy the card, you put your time into testing them further in the interest of getting a good one.  I don't know of many guaranteed "brand" cards failing.  IF a card was going to die, then ML would be the thing you'd want to throw at it.  You're writing as fast as the camera can, generating heat, etc, and putting all the memory cells through max read/writes.  I just don't get it.  To ask if ML is somehow busting something that is a factory second?  Yes, it was a bad card along, as all of them have the potential to be discovered as such.

Interesting to know that.  Out of interest, I googled around a bit to find more info on this information, but the only reference to "factory seconds" and komputerbay was in posts by you.  I didn't look that hard, admittedly.

Would you mind sharing your source for this information?   I think it would be interesting given the volume of use of these cards for raw by people using these forums.

I'll probably get some hate for posting this, but here it is anyway. A comparison between the 50D and the Blackmagic Pocket Cinema Camera RAW.

50D is still the incredible useful for it's price, but if you like the flexibility of the 5D Mark III RAW video but don't want to pay $3,000+, consider the Pocket Cam. Especially since it's now in stock.

Thanks for this, very interesting.  It does kind of reveal what we already know, ie that the BMPCC has much more dynamic range.  I look forward to other tests comparing its raw mode for detail, low light capabilities and moire (also noticed there was more moire on the car grill vs BMCCC compressed)

I also look forward to the real shoot out - 5dII with Dual ISO on vs the BMPCC

OK, didn't know that about ACR as I've never used it.  Sounds to me like you've optimised your workflow.

I use Aftershot pro for raw conversion, but sadly it won't open the DNGs produced by the tools I've tried.

I did get a satisfactory workflow using raw2dng -> DNG -> ufraw-batch -> PNG -> virtualdub -> Cineform 10bit -> Vegas 12 for edit and grade.  What was nice about that was that the denoising in Ufraw was good, but it was slow, and I obviously had to grade on 10bit footage rather than raw.

I have now settled on a workflow based on raw2cdng -> Davinci Resolve -> proxies and round tripping with Vegas (grading after editing for a more joined up approach). 

Have you considered Resolve?

[set the sharpening to 0]

Great work, thanks! I'll be one of alpha testers for this MLV stuff.

A question: does each frame include things like aperture stop and focal length? This is critical, as that's what lens correction module of ACR needs to get read of those annoying aberrations, distortions and vignetting. (Also, having the lens model would be very helpful; but if not, one could always add it later with exiftools.)

I am trying to come up with a "photographer-friendly" workflow for 50D RAW video PP. Specifically, I want it to be built around Adobe Camera Raw (packaged with PS). It actually works very well, except for one deficiency: my version of ACR (6.x) doesn't provide the upscaling option to get from uncropped video to 1920x1080. (It does provide a few sizes, but no way to customize that - weird!). If not for that, one could have a complete solution in ACR - it'd do all the grading, lens and perspective  corrections, then upscale to 1080p and sharpen the frames. No such luck - the last two steps (upscaling and sharpening) would have to be done elsewhere.

Virtualdub is probably the only video tool I know well (have been using it for ages), so I was trying to use it yesterday to supplement ACR. So the workflow is:

• Convert RAW to DNGs
• Load DNGs into ACR, do the corrections, but set the sharpening to 0 - I think this is important, as any sharpening (even the default one) has to be reserved until the end, after upscaling - am I right here?
• Save the frames in ACR as JPEGs - this seems to be the only way to directly bring them to Virtualdub (it doesn't know TIFFs).
• Open the JPEGs in Virtualdub, load two third party filters - Smart resize (I use exact bicubic resize to 1920x1080) and MSU Smart sharpen (I am debating between "Small overall" and "Overall" levels of sharpening), and then save this as an AVI video with the codec of your choice (I use Xvid for now)

At the end, I do see moire/aliasing artifacts, but they seem to be no worse than in the original size video. No flickering AFAIKT. And apparently no new artifacts due to upscaling - e.g. a high contrast diagonal line doesn't appear jagged. I think that's very crucial.

Does this look reasonable?

You'd be better off exporting PNGs than Jpegs.  VirtualDub will use those too, and obviously it's lossless compression vs lossy for Jpegs.

Wow, awesome, thanks Andy600.  I won't have time to play with this now for a couple of days, but will be sure to give some feedback.

Ah well, seems fair really.  Shooting raw on a £250 camera is already a little unfair.  Now, if I sold ALL my gear, plus my road bike and my guitar, could I afford a 5d III? ...hmmm 

Austerity sucks.

Ok cool, well thanks for the FB.  Ain't life a bitch.  My damn 550D does dual iso raw video, but only at postage stamp sizes.  Sounds like a final answer you gave on the 50D, but then again, I've heard final answers before around here

Raw_rec is in the nightlies and aettr.  I don't see mlv_rec in the nightlies, even though it's in the changelog.

I don't know what the edmac channels are, but I get ~75MB/s in Andy600's 50D build, but only ~64MB/s in the latest nightly.

No big deal, happy with what I've got, but looking forward to having aettr with dual_iso support + the speed I'm getting now.  Right now, it's one or the other. 

Playing with mvl_rec would be a bonus.  I'll wait patiently.

Just checked the latest nightly with supposedly mlv_rec.  Alas, module does not show up once camera boots.  Also, ML still doesn't have the raw capabilities that Andy600's TL build has on my 50D

Sorry if this is a boring question, but are the latest features (AETTR seems missing in TL) coming over to TL as they happen?  What are the enhancements that allow such high raw speeds on 50d with TL, and will they ever be pulled back into ML?

Thanks for all the effort to all those who contributed.

I tried on a CR2, but I'm sure you can find workarounds to open a DNG. IIRC, DXO opens the DNGs from cameras that use this format natively, so you probably need to fake the EXIF somehow.

To mangle the data from a CR2, a clumsy hack would be to modify the code from dual ISO preview (since it can burn the preview in the CR2 if you don't call the undo hook). It's nothing more than a proof of concept, not a usable solution. But it proves that it can be done.

I looked into why Aftershot Pro could handle natively generated DNG's but not 'converted' ones.  From what I can gather, both ASP and DXO give up because the metadata is not formatted as expected, specifically MakeNotes or DNGPrivateData.  However, beyond that there is a suggestion that there may be issues with the raw compression for different DNG sources.

I would love to be able to work the DNG's in DXO, because the noise reduction and perspective correction tools are awesome.  Plus it has OpenCL support (as does ASP) which gives a nice boost to your workflow.

Its not that complex but you should be in zoom mode for it to move the box around, otherwise its got nothing to move.

Once digital dolly is on, the joystick does nothing while in zoom mode for me.  If I turn it off again, the joystick moves the focus box as normal.  I understand this is not the intended functionality, but I'm not too bothered about digital dolly until it improves (yes it's feasible I might have a go at coding some improvements if nobody else has plans for it).

I only did a quick test last night with my my card Lexar 32GB 1000x and TL Oct 11 build, and I had no issues recording uncropped 3:2 video at 24 fps, with ~79.5 MB/s. I'll do more testing this weekend.

An idea - sorry if this was already discussed. Is it feasible to create a "virtual rail system" using crop mode on 50D? Are there any technical reasons which would only allow crop video to be sampled from the center of the frame? If not, perhaps such virtual panning is possible to implement? Say, one can specify how many pixels per frame the horizontal shift should be (could be negative), same for vertical shift. Then the crop area would be initially placed at one end of the full frame, and then would shift every frame by the specified X and Y pixels values, until it reaches the opposite end of the frame. I've seen this effect in some movies, looks very cool. I suspect it won't look real when using a wide angle lens (because of the perspective distortions at the edges of the frame), bur for 50mm and longer this should not very very obvious.

You can choose which area is used in crop by moving the box as usual before you record.

There is a digital dolly setting in the raw menu, which is supposed to let you use the joystick to pan the sampled area during cropped recording.  In my case it doesn't work as expected.  Once the option is enabled, the joystick ceases to move the focus box at all, recording or not.  I've looked for documentation on the feature, but obviously it's not in the user guide yet.

From what I've seen demo'ed, it could do with a softer start.  Looks a bit robotic.  Also, would be good to be able to set start and end points and speed as you suggest.

Anyone seen this?  Thinking it might be good for the PI and also for rigging up to a camera:

http://www.kickstarter.com/projects/697708033/hdmipi-affordable-9-high-def-screen-for-the-raspbe

Looks like signing up to the £75 option might just be cheaper than buying the panel and driver yourself, plus you get an enclosure.

9" 1280x800 with just the HDMI input.  Looks ok to me.