Messages

I'm not saying I only use the INFO button.  I'm saying that Canon uses it in their manuals & Canon users as a whole think of it as a button to go to for "info" on the screen. 
I'm suggesting a further change to the UI of the system so we can click the INFO button for a rotation of GD info.

Of course I use GD constantly when doing shots, but right now the only way to turn it on/off is through the ML menu, not with a hard-coded button.

I know AF-ON and Expo lock both have issues, this is why I think we should reroute them to other more useful functions.  Overall, we just have many many buttons on the 50D that could be rerouted and used to our benefits as ML shooters instead of browsing constantly through the huge ML menu.

Some people love timelapse, some love ETTR, some love RAW.  If we restructure the hard-coded buttons so we can toggle on/off what our fave uses of ML is, then it makes our equipment much more useful.  While in stock canon mode, I never take my eye off the viewfinder in my photoshoots.  I don't have that luxury when in ML mode due to lack of hard-coded buttons that I'm capable of using.

The few buttons i mentioned with the ability to reroute to a slew of 12 common functions people use would be a weclomed asset by all, I would think?

2 places, raw module (while recording) and ML menu... I'd say it makes a bit of diff for raw. You're not going to learn all the features in 1 day.

Ok, so we have:

#1 - Overlay screen and a master control at top to toggle Global Draw
#2 - MLV/RAW mode screen
#3 - INFO Button on camera

I suggest we eliminate all but the INFO button as we are used to using this already for overlay controls.  By default, have GD off, then 1st hit of INFO button turns on normal Canon Overlay.  2nd hit of INFO turns on ML GD, 3rd hit brings up secondary ML menu with quick stats, 4th hit completely disables all GD again.

This is one less toggle to have to think about when setting things up.  It is not a matter of whether you can learn it in one day or not, I have had ML installed for years.  The problem lies in so many sub-menus which is not a professional approach.

I personally never touch my AF-ON button, the light bulb button, the Exposure Lock button, nor the Shoot Mode button on my Canon 50d.  Being able to reroute some ML features to utilize these buttons instead of a ML menu with submenus would be incredible.

For example, give us options in the ML menu to reroute these less-used buttons.  Have common options like "GD on/off" "RAW recording on/off"  "Video on/off" and maybe another common that ppl are always using while doing a run-n-gun type shoot with the 50D.  Hard-coded buttons in the config would really make the setup more advanced and help us all out. 

Turn off Global Draw completely and you'll have better results.

As stated in my post, I've switched almost all settings possible and using 2 diff cameras, 2 diff CF cards & giving my feedback on what works best.  Global Draw makes no difference & the UI is confusing for novices since you can change the Global Draw settings in at least 3 diff places.  I'd personally prefer Global Draw hard-coded into the INFO button.  ML is getting crazy with menus & submenus and we really aren't making use of the other buttons we have at our disposal.

It's just not fast enough at 1920x1080. It may do it with the Komputerbay 1050x card . . . But it might also be an issue with the edmacs . . .

The Komputerbay 1000x is definitely fast enough.  I can transfer 117mb/s through my Kingston 3.0 USB CF Card Reader just fine.

I think there are still some limitations in the programming.  Honestly, we only really need to squeeze another 2mb/sec and then we should be good.

In further testing, I switched around a ton of settings and found there may be something eating up memory to control the lenses.  I tried out a few Tamron lenses that my camera doesn't recognize and they gave better results than my L-series or nifty 50 Canon lenses.

While recording, I assume there is something in the program that is reading aperture settings or focal points, even when in manual mode on the lens.  Has anyone else tested their speed in relation to having a generic lens on or not?  I had all lenses in manual mode while testing.

I get continuous at 1920x1038, though taps out around 2 minutes on 1920x1080p.  This is all in 23.976 fps mode. 

It doesn't make sense, though... the buffer is doing fine & then it all of a sudden ends at the 4gb mark like it is a 32-bit related issue as opposed to buffer.

My buffer is only half full and then right at 4gb mark it quickly fills.  I've tested this on both cameras and at various places on each CF card, so I dont' think it's a hiccup in the card either.  Maybe still not fast enough and coincidentally always capping at that same 4gb spot.

So, I have 2 50Ds.  Both using 64gb 1000x KomputerBay cards.

Just installed latest Tragic Lantern & wanted to give some feedback when shooting in cropped mode for 1920x1080p.  This is for both RAW and MLV modes, I get almost exactly the same results.  Card hits 78mb/s in both cameras for ~20 seconds on first run. Then, 2nd run gets up to 82mb/s and sometimes peaks at 83mb/s.  Every file tops out around 4gb-4.4gb 

Also put into test mode and it seems M2 B0 is fastest on both cameras by up to a +4mb/s increase.

This is true when hooking up via HDMI, leaving global draw on, or turning off - none effect performance. 

Also, my LV never gets turned off, even when going into Hacked mode.

While recording, no matter what mode I'm in, the screen turns grey once I hit SET for recording.

Hope this helps! 

  In windowed (5x/crop) mode, it's 3584x1320, but it's unlikely that recording the maximum resolution will ever be possible because the card writer in the camera is limited to 160-170MB/s and that size requires a bit more than that (190 MB/s).

Is it confirmed that the card writer in cameras is limited to ~170mb/s?  If so, that is the main bottleneck we face & surpassing the limit would mean a hardware hack to bypass this card reader straight into a diffferent storage system, such as SSD.

It's coming but it's a lot of work and I have to write it in between other work. I have about 40 pages written so it won't be too much longer

Why not make a wiki or even a Wordpress page with free access for others to help out?

I designed an app that allows this easily if you want an account at www.openbuk.com - I'll contribute a free account for this use if the community wants one. Just lmk

Here's my latest footage of a Bikini Commercial using 50D Raw with June 28th build.  I'm using a 16GB SanDisk Extreme Cards as I knew I wouldn't be shooting more than a few seconds at once.  It worked out well. Edited in After Effects. Lens: Tamron F2.8 17-50mm + Canon f1.8 50mm.

One thing to know:I picked up the Komputer Bay UDMA7 card that is 128G.  Could this be the problem? Maybe the it should be 64 or maybe UDMA6?

The 64GB Komputer Bay is faster than the 128.  Also, it is hit or miss with each card, so you can return it if it doesn't do well, however, settings like GlobalDraw & others may be interfering with your speed, too.  Rule of thumb - turn everything off before recording.  Hope this helps.

The jpeg function still didn't work for me when I was tethered, just crashed LV. On 600D tetherin worked... seems like a nice idea to revisit and possible to just cache hack the jpeg to be larger sized... it has the whole raw size to work with. But even in EOS utility the FPS is closer to 20 with the small ones.. maybe the jpeg engine isn't fast enough? And 2 in 2 out would imply it processes 2 frames at once?

Also lv_save_sndpass size should be investigated for 422. That way maybe there could be a choice, RAW -> 422 -> jpeg. Would completely obsolete H264

Yes, this is why I want to revisit it.  I believe the knowledge gained from RAW can be applied to what you have already tried on JPG in past. 
I feel the architecture of the canon firmware was designed to accept slow CF cards.  We could optimize this for 1000x cards only - which would cause slower cards to crash, but at least give us new buffers & capabilities for the faster cards we have today. 

The RAW -> 422 -> jpeg setup sounds great as long as the buffer doesn't get filled too quickly since we will be re-introducing compression.

That's interesting. 14-bit RAW videos with a resolution of 4096 X 2304 at 24 frames per second require 378 Megabyte/s write speed.
Might become a little tight with the 50D's write speed ...

But for silent pictures 4K would be a nice feature on all Canon DSLR.

I'm not talking about 4K raw.  I'm talking about JPG at 4k.  This requires significantly less throughput.  We already have small JPG image capability on the camera, just need to put this in silent burst mode and see how many frames per second we can make happen with it. 

This mod will allow us to sacrifice dynamic range & low-light capability for increased resolution with the same bandwidth.  We can also start with 8-bit and work our way up to higher br until bandwidth is at max threshold.

Yeah contact @Gregoryofmanhattan his been working diligently on 50D's code & thanks to his hard work we currently getting 70mb/s.  Are you talking about 4K build for the 50D or I'm reading it wrong? Is that even possible?

I just forked @Gregoryofmanhattan on bitbucket.  Hopefully that will work.  I am talking about 4k on 50D.  It will definitely work, it's just a matter of how high of quality it will be & if reverse-engineering is worth the effort.  I'll post progress on bitbucket. Best of luck to other weekend coders!

RED recently released a new 4K conversion to make .RED files instead of R3D.  I know that is a long shot, but we could potentially override the compression agent of Canon and replace it with RED conversions.  Supposedly, this records 4K footage at 2.5mb/s. 

Regardless of that accuracy, RED has been in the RAW business quite awhile and if we either use their compression tactics or switch over to using H.265 - we can really improve the quality within the limited bandwidth that we have.

Has anyone tried porting H.265 or .RED compression already into ML?

Thanks for the insight - exactly what I was looking for.

I don't want to try a direction you have already put tons of hours into.  We essentially need to find a solution nobody has thought of yet.  Surely, it can be done if we can override shutter getting used & make the resolution small enough to fit within ~96mb/s.

Since I'm new on your mod, It will surely take me some time to do this... though, I see it as a better investment to help this community than to purchase new hardware that is capable of this.  If we build this thread with all our trials & errors - the future programmers can learn from our mistakes and finally get this sucker finished.

I'm working on a 4k Build and other tweaks... just wondering where the latest branch is which I should use for the 50D?
I know multiple ppl have various branches.  Any insight on which has the cleanest & most up to date code?

One method that I see working:

Add a new JPG Photo mode that saves photos to exactly 4k frame size.  Then, use burst firing mode to see how far we can push the buffer.
If this doesn't work, create a photo mode with the highest available burst firing mode possible without exceeding buffer.

We could also code today and plan for tomorrow.  Go ahead and write code based on the fact that UDMA 7 1666x speed cards will be available within a year that can provide this threshold.  Or, in the meantime someone may figure out the CF to SSD converter process. 

Either way, finding a solution to provide the menu options should be our first focus & then worry about buffers and such thereafter.

I only have a Canon 50D out of my lineup that can potentially handle this, so I'll play around this next week and see what I find.

Maybe I didn't make myself clear in my last message.
We need a thread specific for 4k RAW.  I'm a programmer & I'm not asking ML to put this together. I'm asking the programmers who have already started on this to actually start a thread & we all share our input to the code that we have seen.

I can contribute time & energy on programming to this.  Though, I'd prefer to see data from multiple programmers all in one inclusive place while I put my effort forward.  It doesn't make sense to reinvent the wheel - as any programmer knows.  We are lazy & we use the most efficient method possible to start. I'm not seeing positive community involvement in locking my thread when I'm putting effort into an area that has no substance as of yet.

Great success on the RAW quality everyone!
On a different note, there are filmmakers such as myself who have different interests.

I shoot with 3 cameras at a time for multiple angles on my footage. I would like to stay ahead of the curve and offer 4k footage as opposed to 1080p. If you are smart about lighting your shots, you can get away with using jpg quality footage just fine.
However, we currently don't have a way to perpetuate our footage into the 4k market.

Pushing CF cards to a 100mb threshold should allow us to barely squeeze by with 4000x2000 resolution jpgs at 4mb each. 4mb per photo X 24 photos per second = 96mbps. This is essentially 4k non-raw footage.

Is anyone currently working on this? I've learned to master my lighting techniques to makeup for lack of RAW over the years, though focus on 4k will really give our cameras new life for literally 10 more years.

I see there has been some chatter about the topic, yet the new knowledge with how RAW video works should give us insight to work on this direction as well. Anyone else have this desire for their film needs?

Off topic I know, but does anyone have an external battery solution for the 50D? I was looking to buy this - http://goo.gl/to1vd - but I believe the 50D uses different batteries and so it may not work. Any ideas?

There is a solution for the 50D.  The Canon DS8111.  I bought mine at Samy's Camera.  Here's one on ebay.

Just like Bokia mentioned, you can get multiple batteries that are high mAh.  I have 4 2200 mAh batteries that literally last 3 hours.  Use them in a battery grip & you will be set.  I've shot on 12 hour productions with these out in the field over past 2 years with great results.

I'd rather we all work on a team on this.  I'm a community member just  like you & helping out ML.  I'll talk with my lawyers and see if they have any similar scenarios.  Maybe someone else on here can find something or knows of a company that does something similar?  Let's see what we can find.  I totally understand that concern.

[magiclantern.com - not .fm]

Yes, I am a businessman who enjoys helping charities & philanthropic efforts.  ML has a great cause & great people behind it, yet still a weak foundation for the business that it is.  It is quite common for such a company to get $500k funded literally overnight when the right business structure is setup to it.  I haven't seen other contributors here who are programmers in the Silicon Valley area.  Maybe that is why everyone thinks it has to be garage style to still be "for the people."

If you look at other large & successful open-source projects, they handle their work slightly differently. Think wordpress, eclipse, kaltura, android, chrome OS, etc. These are all open-source yet still bring in funds to help their beloved programmers & their vast communities grow.

Here are my suggestions:

#1: magiclantern.com - not .fm - this will be the for-profit side of the business where people can hire programmers to help them install magic lantern in all of their corporation's cameras, etc. 
have magiclantern.org - which is the developmental side of ML and accepts non-profit donations to further educate people how to use the software & continues to contribute code
(this setup is exactly how kaltura.com works, wordpress.com, etc - many large open-source programs)

#2: Get a board of directors & rectify a nonprofit.   anyone on the board will not be allowed to accept any money - they simply oversee decisions & stand behind the mission of the nonprofit.  elect a board who has ML's best interest in mind who have diverse talents such as programmers, filmmakers, marketers, and businessman.

#3: Do outreach with ML and partner with Canon.  Once ML becomes a 501c3 nonprofit, Canon can make tax-deductible donations to the ML crew.  Also, I'm sure Canon would be happy to hire programmers from the ML crew & can create a happy cohesive environment between ML + Canon.  ML can be a training-ground where anyone can learn Canon firmware & if they really excel, they get picked up by the R&D department of Canon.  With $500k funding, non profit status, and a board of directors that sets this structure up better - Canon will happily start doing business with & welcoming the ML community a lot more.

#4: Write in your mission statement that the purpose is to explore what cameras can do for 100% free & share that knowledge with the world. 

Never mistaken someone like me who gives sound business advice as trying to "take money".  You're welcome to see my business & ventures to realize most of what I do is philanthropic & has the heart behind the grassroots community in what I already do.  I'm merely suggesting ideas and hope to see the ML community latch on to them and make them part of the whole.  Simply a seed planter

If we put Magic Lantern on the MIT open-source software license or something similar and make sure legally it is 100% open-soucre & also 100% just for research and education - then that allows us to freely modify & raise funding for the code.

I already have lawyers on my staff and this is how we develop other open-source software.

I didn't look through the code thoroughly, but every single page needs to be on the MIT license for this to work accurately.  A simple setup is to have people sign in the TOS when they signup for ML website that they comply to doing this for research only & that any modifications they make must also comply to the MIT license.  This can also be considered a GPLv3 License, which allows the same thing.

The only problem I see is if someone modifies ML coding & decides that their modification won't go on the MIT or GPLv3 license.  To circumvent this, we can put a clause saying all future modifications MUST also use the same license.  This will keep the software free forever.

They're already doing what they're doing and kicking butt at it!

They have done a great job up until this point, though it is still an amateur piece of code with inadequate licensing and maintenance to really hold standard in a larger marketplace.  If we treat it similar to the Android system, we could theoretically have a platform for future chipsets to run only on MagicLantern Firmware.  This would allow a complete different open-source camera system to adopt ML into it's hardware for a truly 100% open-source platform.  This can't happen on the random hours the current coders are putting into this software.

Whether we like it or not, writing code takes money.  This doesn't have to be a for-profit organization in order to fund the cause. 

A few areas funding could go to:

Dedicated Servers to host ML website: $2,000/year
Redesign responsive website: $4,000
How-To Tutorials for beginners: $10,000
Funding for 5 programmers to treat this as part-time job: $50,000 = $10k/year per programmer. 
Distribute the $50k to the top 5 programmers who work on this code which will allow them to buy cameras and completely take them apart, completely obliterate CF/SSD controllers, etc.

We need to step away from the mindset that ML is a garage/basement project and realize that it is currently ranked at 59,000 on Alexa, has global coverage with top magazines and reviewers, and is being used by filmmakers worldwide in commercial productions.

We're essentially sitting on a multi-million nonprofit educational open-source platform that just needs the right management to standardize how it works. 

Adding commercial advertising on the website, kickstarter projects, and writing books on how to use the software are just a few commercial aspects of financing that can be done without making any of the actual software profitable or closed-source.

Let's rethink this entity & someone with more a higher post count than me take over the movement in that direction

Magic lantern does take donations and even blocks out the download of software and demands users pay $5 to get the build before anyone else.
That is a commercial business and we can step away from that mold by offering other forms of donations and funding which will allow us to make magic lantern a larger and more solid platform.

In a strong company, you have multiple marketing campaigns raising money in different directions that all go back to the main source. So regardless if Ml gathers funds elsewhere, gathering them on a new site seems appropriate.

I mention kickstarter since the energy on there is very welcoming to the open source community.

I work with open - source software for a living and in looking through the Ml coding, I see the programming could be taken to the next level with some professional programmers.  Even training the current programmers or providing them funds so they can do this full time instead of offshoots from their other jobs or from their schooling.

My interest is in expanding ML and expanding awareness so we can tackle the challenges in firmware with sharper minds, more focus, and finding that enables all of this to happen.

I vote for funding what we already have and taking ML to the next level.  Add better support here on the site, give the Devs some money for servers & pro bitbucket accounts, etc.

I have opened a thread about this.  If we get funding to support the existing ML firmware, then we could potentially take it a degree further.  Regardless, we as a team need to see this project as fruitful and put energy into that direction.