Guessing 1st April for 2019 - Share your thoughts

[dfort]

...I've started to question why I'm even doing this.

I've been wondering that myself too. Yet somehow playing around with Magic Lantern is too much fun to let go.

I need to change something.

Sounds like you have some good ideas. These days so much can be done with software. Years ago when I was getting started in photography I went to New York City to do my apprenticeship. There was this one legendary camera repairman, Marty Forscher, who ran a business named, Professional Camera Repair. He did hardware hacks like grafting Polaroid backs onto Nikon F's, amazing stuff for that time.

Anyway, just thought I'd throw that out there.

[andy kh]

If someone interested (Making MLVApp x10 faster) let him take this task and find a way where to collect the money from interested users and give guarantee to not steal our money haha , we need a trusted guy.

Then we can save a lot of time for processing and push the things forward.

I can start with 5$ .

Isn't it easy?

im in

[garry23]

I'm not going to add to this in any direct way, as it resolves around how people feel and their family/life commitments.

However, from my perspective, it was rather a coincidence that I was at a meeting yesterday when someone made an interesting statement.

In the past software came after hardware. That is software was always being changed to keep up with hardware changes. I guess the ML community has seen this with the changes that are required each time a new camera is released.

What some are saying now, is that software needs to be recognized as being their first, i.e. developed on virtual hardware, and real hardware is adapted to software.

I'm sure others know more about the above potential futures than I do, I just thought it an interesting reflection.

[Luther]

I've been using ML for years now. This forum and ML is what made me go to university study photography. Many of you have helped me through those years, in special a1ex and g3gg0, but also many other devs and users.
@a1ex, just wanted to say ML isn't for nothing. I truly think you have made a revolution with your development on ML. You gave people a tool that can record in a quality that once was only available in high-priced cameras. You gave people without money the possibility to compete in quality with high-end cinema, the possibility to express more freely. This isn't "nothing". I'm grateful to have this forum and ML.

As for the money, I agree with @Ilia3101 , a Patreon page could make it easier for people to donate.
Other idea I just had was to make a short film, just like Blender Foundation does. Could be a community effort and all the money gained through it can go to ML development.

I will discuss this topic in the forum in the right time and will let you know all the details, if someone want to help it's will be cool too.

About MLVApp, maybe a government funding? Natron got funded by INRIA and it seems to have worked well for them. I don't think this is an option for ML (because of copyright stuff), but MLVApp would have no problem trying.

[IDA_ML]

Hello A1ex,

I am not sure if you will find my thoughts useful but since you said that you welcome other users' suggestions, I decided to write them down.  I don't know you personally, don't even know where you are from, that is why I am not in the position to tell you what to do but I will tell you what I would do if I was in your shoes.

In fact, I am in a situation very similar to yours.  The only difference is that I do not work on such huge projects like ML or Apertus.  I have a full time job that is very miserably paid and about 10 years ago I decided that I should do also a second job to cover the basic needs of my family.  I decided to turn my beloved hobby into a second job and became a freelance photographer.  I was doing this job on weekends, holidays and vacations and in the beginning, I was quite successful.  At that time I was even thinking to quit my full time job and try to make a living as a freelance photographer only.  God, am I happy that I didn't do that because if I did, today, this would have turned into a financial disaster for me and my family.  The reason is that shortly after that successful period, photography started declining very rapidly and nowadays, it is extremely difficult to make a living as a photographer, in my country, at least.  I remember a professional wedding photographer saying recently:

"There are more wedding photographers than weddings these days." 

The bottom line is:  If I were you, I would never quit my full time job, especially given the fact that, as you said, you have a growing family to support.

Now the question is how to find time for ML and also other projects that you find interesting and rewarding.  Well, I would just keep working on them in my spare time, just as I do right now with several activities that I find interesting and that I am fond of.  I would just do it in a different way:

1) I would make small steps, one at a time, and wouldn't start something else before I am done with the previous one.  You mentioned the crop-recording module.  Just go ahead and work on it until it does what it is supposed to do and you are finally satisfied with how it works.  Then release it for testing.  In this way, you would keep the testers busy, will make the users happy and will receive a lot of feedback on how to make it even better. 

2) I would do my very best to try and keep the interest in ML alive.  A few years ago, the interest in ML among photo and video shooters in my country was quite high and many people started experimenting with their ML capable cameras which, at that time, had lots of flaws and short comings.   All these people waited for improvements but they waited too long.  Finally, they became frustrated, sold their Canon gear and most of them transferred to other camera brands.  If all these exciting new functions such as crop recording, card overclocking, lossless compression, sensor readout methods, ISO research, etc. had been updated more regularly, this would have kept people's interest in ML alive and they would have stayed with Canon and ML. 

You have now released the ML port for the M50.  With all my respect to you, I think this was a wrong decision.  Why?  On one hand, this is a new model and very few people have it.  It is not an established model and it will take time until people start buying it.  Until then, it will be quite expensive.  On the other hand, as you said, it is quite different from the classical ML ports and you probably spent a lot of time with that port.  Moreover, I guess, you managed to implement only the basic ML functionality into the M50, so until it gets all the functionality of older but much cheaper cameras like the EOS-M or the 700D, there will be a long way to go.  If you had spent that same time to port ML to the 5D4 or the 80D, this would have been a real break through in the forum.  There are so many users shooting these cameras and they all have been waiting for years to get their hands on a working build for them.  Imagine how many people and other developers would have jumped in and helped with the development.  I am sure, many users who had these expectations, are quite disappointed now.

3) There are many knowledgeable and skilled people on this forum who do a lot of useful work and help ML with new developments.  Very often, these people come to a point where they badly need some help to go ahead with their work and only you can provide it.  One example is Dfort's work on the 7D.  He spent years of his time and a lot of efforts trying to implement 10/12-bit and crop recording to the 7D.  At a certain point, he simply reached the limit of his abilities and had to stop his work.  As a result of that, the 7D development stopped for nearly 2,5 years and is not very likely to continue without some help from you.  You don't have to do much to provide this help.  Sometimes, just a few hints in the right direction can help a lot.  Remember what happened with the 5D2?  Reddeercity was really very persistent with his efforts to get this camera working in the high-resolution modes but the real break through in his work happened only after he received a few key hints that pointed him into the right direction.

So, if I were you, I would work in a closer collaboration with the other developers, follow their work closely and provide the help that they need to keep going.  In my opinion, this will make their work much more efficient and will take a huge load off your shoulders.

There are also so many new enthusiastic and highly motivated users on the forum that are offering their help all the time.  Some of them have the required technical background and coding skills and their help would be very useful.  Why not simply contact these people personally, see what their motivation, interests and skills are and assign some of the easier tasks to them?  In this way, you will have more time to work on the serious work that nobody but you can do.

To be continued.

[Lars Steenhoff]

The main problem is you can't please everyone, I mean its not a really a problem just something to be aware of.
And its great work who all that contribute to magic lantern have done, There is no single best way in my opinion, the learnings of each little step contribute to the overall

[Danne]

@IDA_ML
Take a look at what is added into Magic Lantern repo:
https://bitbucket.org/hudson/magic-lantern/commits/all

Then check latest work that has been going on and people included putting in efforts here and there. If you read carefully you´ll see the holistic approach.
I develop(well more like refining ongoing code, I would never take credit for the completely encryptic low level coding achievements taken years for a1ex, g3gg0 and a few others to accomplish), I do not "need" help. I search until I find an answer, sometimes I get a hint and even deeper help, as to when taking a fools step into running prop regs at the wrong places  . Thing is. All code is out there. Taking baby steps, if so even just starting to stare at the code will be more rewarding than dreaming of others to do your work.
Besides that. Personally, I look at the m50 as the perfect successor to the eosm. Just perfect.

[andy kh]

there is nothing wrong porting ML in m50. there are so many m50 users waiting for this moment and it's the perfect camera as it is not too expensive and much cheaper than a used 5D mark III. i am eager to see how to beast turns out... i wil also buy one if it can shoot some good slowmo without moire/aliasing like 5d mark III

[jackmoro]

You have now released the ML port for the M50.  With all my respect to you, I think this was a wrong decision.  Why?  On one hand, this is a new model and very few people have it.  It is not an established model and it will take time until people start buying it.  Until then, it will be quite expensive. 

Quite wrong statement. ANY new camera port is great. I wish it was 5DIV, but when M50 will be running MLRAW I will buy it next day and use it until there's R, IV released. So it looks like quite personal for you.

[masc]

I, for one, would pay $10 for an Android app that can convert dual-iso files. 

Why not just installing android sdk and Qt and start to compile MLVApp for Android? Qt is cross platform and supports android too. Maybe a lot of Desktop stuff has to be removed and exchanged, but I don't think it is impossible. It just will be 100x slower as on a PC, because of smartphone processors.

Also because Magic Lantern has helped me a lot , MLVApp too , I have started to search to a Cuda programmers for making MLVApp GPU Accelerated.

GPU acceleration would be great. But why CUDA? This works on nearly no computer, at least on none of the >10 computers I have around me. OpenCL or OpenGL works nearly everywhere nowadays...

@a1ex and other devs: I know what you are writing about. All the topics here, a full time job, a family and other hobbys... and in the end the day has just 24 hours. It is very difficult to find a good balance. I feel it nearly every day... ... but I am sorry to say, I have no solution which works for everybody.
See my part helping with MLVApp as thank and respect for your work over all the years with Magic Lantern. For me this project is something really great!

[andy kh]

GPU acceleration would be great. But why CUDA?

i have a 10 yr old laptop with cuda. it can play cdng in resolve in almost real time. it also plays in realtime and very powerful when i edit in premiere pro. i never needed to convert to mov or proress to edit in premiere since cdng can be played very smoothly in realtime in premiere pro. this is my experience wit cuda and i know who use cuda will wil love cuda as much as i do

[a1ex]

You mentioned the crop-recording module.  Just go ahead and work on it until it does what it is supposed to do and you are finally satisfied with how it works.

During this summer / autumn I've been researching the image capture process for quite a few months, but... at some point I had to take a break. All these FPS timer values were effectively spinning in my head, and I felt the need to focus on something else to clear up my mind.

The crop recording module was announced in 2016. How much I should keep working on it until I'm allowed to look into other areas of ML?

Another example. About one month ago I've found a bug in QEMU, while helping chris_overseas with 5D4. The bug: it misses some of the DryOS task switches, when logging function calls ("-d calls", if you've already read the QEMU docs). It's a feature I use pretty much every day for development. OK, I've started to fix it, got it somewhat working and ran it through the test suite. Still crashing, or giving wrong results, on some models (mostly on DIGIC 7 and 8 ). I kept wrangling the code for about one week (March 5-9, from my local history). I happened to have a small holiday at that time, so I've effectively wasted it on troubleshooting that bug. It did not result in a commit yet, as I was still not happy with how it worked. How long should I keep trying before moving to something else?

And it's not the only one. I frequently get into rabbit holes with things that don't work. This is the nature of reverse engineering - you don't know in advance whether the approach will work, or not. The few things that I manage to get working, are always published here on the forum. The ones that I can't get to work... well... they are moved to the back burner and I revisit them after some weeks or months (possibly armed with new knowledge).

If you had spent that same time to port ML to the 5D4 or the 80D, this would have been a real break through in the forum. 

Just FYI, I've been focusing on pretty much all DIGIC 6/7/8 models in the last months, maybe more. Yes, including 80D, 5D4, 200D, 77D, EOS R, 7D2 (still waiting for test reports there), 5DS/R. Everything I did on the M50 (except maybe less than 10-20 lines of code, which were specific tweaks for PowerShot firmware) is going to work on all the above models, with little or no changes. I expect the "Hello World" demo to be already working (already confirmed on 80D, 5D4, 5DS R, 200D, 77D, besides M50).

I've got the M50 3 weeks ago. I've spent a couple of days implementing some logging code + Hello World, and I've also made sure it works on most other DIGIC 6/7/8 cameras. For Hello World, I've picked at least one model from each generation, and all of the "difficult" ones - commit 80ed550:

Code: [Select]

M platform/200D.101/stubs.S
M platform/5D4.112/stubs.S
M platform/5DS.111/stubs.S
M platform/5DSR.112/stubs.S
M platform/77D.102/stubs.S
M platform/7D2.104/stubs.S
M platform/80D.102/stubs.S
M platform/M50.101/stubs.S
M platform/R.110/stubs.S

The remaining models (not covered in that commit) are very similar to the above, i.e. anyone with basic coding skills and some patience can fill in the gaps.

Pretty much every single change I did for DIGIC 6/7/8 was made with *all* camera models from this family in mind. Yes, this is slowing me down, but this is to make sure all ML users will be able to use my tweaks eventually, and I believe this approach will be helpful in the long run (as opposed to picking one model and focusing on just that).

Yes, this means I've neglected crop_rec, for example. Or 1300D. Or many other topics. I know that. My internal task switching routine is quite slow; once I start looking into something, I tend to keep doing that for a very long time until I give up. And when I give up, it takes a while until I revisit the original topic.

As a result of that, the 7D development stopped for nearly 2,5 years and is not very likely to continue without some help from you.  You don't have to do much to provide this help.

Not much, only a couple of months of full-time work (or maybe more). 7D is not your average DIGIC 4 camera; I'm pretty sure I've answered it before. You seem to know the difficulty of this task in advance; why don't you do it?

If all these exciting new functions such as crop recording, card overclocking, lossless compression, sensor readout methods, ISO research, etc. had been updated more regularly, this would have kept people's interest in ML alive and they would have stayed with Canon and ML.

What do you think I'm trying to achieve here? Exactly - finding the time to do all of that.

See my part helping with MLVApp as thank and respect for your work over all the years with Magic Lantern. For me this project is something really great!

Thanks Masc! Yes, I appreciate MLVApp a lot, even though I don't edit video myself (and that's because I find it too time-consuming).

[ArcziPL]

The only difference is that I do not work on such huge projects like ML or Apertus.  I have a full time job that is very miserably paid and about 10 years ago I decided that I should do also a second job to cover the basic needs of my family.  I decided to turn my beloved hobby into a second job and became a freelance photographer.  I was doing this job on weekends, holidays and vacations and in the beginning, I was quite successful.  At that time I was even thinking to quit my full time job and try to make a living as a freelance photographer only.  God, am I happy that I didn't do that because if I did, today, this would have turned into a financial disaster for me and my family.

If we are about giving life advices... Man, 10 years at very miserably paid work? I don't think staying there is a good decision at all. So far I've changed my employer twice in pretty short carrier. First change resulted in immediate salary increase x3, much less stress and much more free time while doing comparable job, second change gave immediate +50% at absolutely identical job.

"There are more wedding photographers than weddings these days." 

But the best ones are fully booked one year upfront, easily charging 3x-4x more than the average ones.

The bottom line is:  If I were you, I would never quit my full time job, especially given the fact that, as you said, you have a growing family to support.

Well, it all depends on alternatives. But I doubt ML can be such alternative. How much was donated in 2018 and 2019? Yeah, a1ex mentioned... there were two donations.

You have now released the ML port for the M50.  With all my respect to you, I think this was a wrong decision.  Why?  On one hand, this is a new model and very few people have it. It is not an established model and it will take time until people start buying it. Until then, it will be quite expensive.

Your arguments are invalid, sorry. According to some news, it was best selling interchangable lens camera in Japan in 2018. M50 is also damn cheap. I got a new M50 with 15-45 and 55-200 from official retailer for 100€ less than just 80D body (and plan to sell these lenses, getting 300€ back, so the body with cost a half of 80D). M50 and 80D have the same sensor. And with M50 I can finally forget all this AF front- and back-focus shit, which drives me crazy since my first DSLR. It also gives a choice: taking lightweight with EF-M lenses and having great performing cam, what I can't really say about EOS M, or all my damn good but damn heavy glasses (Sigma 18-35 f/1.8 Art, Canon 17-55 f/2.8, Sigma 50mm f/1.4 Art, Canon 85mm f/1.8 ), which I love for IQ but get pain in my back if need to carry them for longer. One cam for all cases. As a result: I don't want a mirror anymore. Ever.
And I am totally for a new, recent model with an up-to-date sensor (!) than spending time for any 10-year old cam, which will only loose importance with time.

On the other hand, as you said, it is quite different from the classical ML ports and you probably spent a lot of time with that port.

According to a1ex they are all similar and much can be reused. M50 just needed more workarounds, so was harder and actually starting with M50 is IMO a good decision. Or same good like the others.

If you had spent that same time to port ML to the 5D4 or the 80D, this would have been a real break through in the forum.

I personally don't see why 5D4 or 80D would be such breakthrough, compared to 5D3 and 70D. Quite the opposite. M50 is a cam like no other supported by ML.

There are so many users shooting these cameras and they all have been waiting for years to get their hands on a working build for them.  Imagine how many people and other developers would have jumped in and helped with the development.

As much as now: single persons or none. They (& we all) had a possibility to jump in and now have it even better. I don't see crowds. 70D, so far the only cam with revolutionary DPAF in ML support did not achieve "fully ported" state after 6 years on the market. Why should it be different with 80D?

I am sure, many users who had these expectations, are quite disappointed now.

Here we come to the clue. Is a1ex's goal doing good in his spare time to all people, including the forum trolls just asking every week "when it will be done?!?!?!?!!1111 I waaaaaaaaaaaaant!1111" or just having fun and satisfaction doing what he finds interesting, challenging, useful and then sharing it with others? If I was him, such moaning and groaning would totally demotivate me from any contribution. Really. His job is great, it's wonderful that he is such a passionate and shares results of his work with us. Whatever he does -- take it as a gift. You take what is given to you or you get nothing instead. Then only letters to Santa Claus are left.

[theBilalFakhouri]

GPU acceleration would be great. But why CUDA? This works on nearly no computer, at least on none of the >10 computers I have around me. OpenCL or OpenGL works nearly everywhere nowadays...

It was just an example  and because it's supports multiple GPUs unlike OpenCL and not sure about OpenGL , but anyway I don't have Nvidia SLI , OpenCL is the best choice yes, I will make a search of these things with you men and decide together, maybe the both.

See my part helping with MLVApp as thank and respect for your work over all the years with Magic Lantern. For me this project is something really great!

My part will be on youtube soon. ML needs to spread more.

[dfort]

Saw my name in print so I had to read it--the whole thing!

...One example is Dfort's work on the 7D.  He spent years of his time and a lot of efforts trying to implement 10/12-bit and crop recording to the 7D.  At a certain point, he simply reached the limit of his abilities and had to stop his work.  As a result of that, the 7D development stopped for nearly 2,5 years and is not very likely to continue without some help from you.

Ouch, that hurts and it is mostly fake news. In the past couple of years since I got a 7D I've played around with doing a firmware update which was no small task because at the time I had to use the dot-dumper. I tried a few times to get 10/12bit working and after a few false starts I finally got it working thanks to a1ex's help. Note that this happened just one month ago--not 2.5 years!

You don't have to do much to provide this help.  Sometimes, just a few hints in the right direction can help a lot. 

Give a man a fish and he eats for a day, show a man how to fish and he'll sit in a boat all day long drinking beer. Or--in this case it is more a matter of taking the hint and figuring out how to work with it. Sometimes it is hard but it is much more satisfying than having everything handed to you--then complaining about it.

To be continued.

No!!!!! If you are going to continue please do so in a positive manner. If Steve Jobs would have stayed working at Atari, if Bill Gates hadn't of dropped out of Harvard and well, if I would have listened to all the naysayers that told me that it is impossible to get into the movie industry---

IN THE END… We only regret the chances we didn’t take, the relationships we were afraid to have, and the decisions we waited too long to make.

― Lewis Carroll

Go for it Alex!

Oh--and good choice going with the M50. Currently one of the most popular cameras on the market, Digic 8 processor, mirrorless and most important of all it is a camera that is still being manufactured.

[theBilalFakhouri]

I think a1ex made the history for porting DIGIC 4 & 5 Cameras to ML and He made the future by porting DIGIC 8/7/6 Cameras which will becomes a second history after ten years.

@IDA_ML
Maybe you was looking forward at 1st April for 4K in 7D, and 1x3 Binning for 100D , if so try to do it it's not very hard. I could do it if I have the two cameras (I am thinking to buy 600D to play with it too) I am someone like you not experienced in programming and compiling and also electronics but these things are easy like drinking water if you learned have to use it.

Okay EOS M50 will have less crop factor than x2.5 in 4K mode from Canon panel by using RAW video by Magic Lantern, maybe 2.0 or less (It will shoot more than 4K maybe 5K) . Just I am guessing the future.

One day a1ex will say that's enough for me I made a lot, and he will write an emotional message (I am the first one will cry on) to all other developers and the community, we love a1ex but a1ex isn't forever, Documenting things is our hints in future to me and to the next a1ex and other users and developers, I am saying this because we should care about guiding , the best way is to make it in videos and to explain everything from the beginning how the internals works in the cameras etc . .

All the Solutions we are looking for and other stuff can't be happen if didn't understand how things works. As I said I am planning to a youtube channel for this, making things quicker, easier and for everyone.

Hats off Magic Lantern.

[garry23]

As I've been reading the above posts and reflecting on the 'ML journey' todate, I believe there are three broad themes that remain in play:
 a. Continuing to get the best out of the 'image side' of our Cams
 b. Continuing to get the best out of the 'video side' of our Cams
 c. Using Lua scripting to get the best out of the 'workflow' when using our Cams

I've personally never been 'interested' in b., but fully recognise why some are. In fact, these days video appears the number one topic, by far.

From an image perspective, I believe, other than correcting errors, 'we are there'. That is we have Raw-Spotmeter, Raw-ETTR, Auto-bracketing, Dual-ISO etc.

For me, the area that I would love to see taken a little further, and to date this has been done by a1ex, is Lua scripting, ie opening up more functionality and finding solutions to a few remaining niggles, eg focus control; as this allows the ultimate enhancement to our Cam, ie making them work, individually for us.

Playing with the core of ML requires real knowledge, maybe further fixes to Lua can be accomplished by others. Sadly not myself, ie I'm only a Lua user.

As I say, just some personal thoughts, especially directed towards one of the killer features in ML, ie Lua scripting.

Bottom line: like many, I'm proud of the ML community, but also like others I recognize ML is built on the shoulders of giants.

[a1ex]

4K in 7D, and 1x3 Binning for 100D , if so try to do it it's not very hard.

On 7D, I'm afraid it is. I don't have a quick solution there; it needs to be done on the master processor, but ML is currently running on the slave one. Figuring out how to run ML on master, without publishing the FIR encryption tools (cough DMCA), is non-trivial, and likely just the first step for resolution overriding; much like Hello World on a DIGIC 6/7/8 camera, vs a full ML port.

Yes, we had attempts to figure this out, together with g3gg0, and I can document the unsuccessful attempts if anyone thinks it's of any use. I believe it will end up as many other things I've documented - nobody else being able to use them. It is really an advanced topic.

We only got as far as emulating a tiny part of the IPC communication in the bootloader (but that was not enough to cover the main firmware startup). That doesn't count as "Hello World"; that's more like being able to blink the LED - to get the idea. The last log files I've got are timestamped December 2017, and I remember trying for 1-2 weeks or so of focused work.

On 100D, and all other single-digic 4/5, I agree, it's not very hard.

[Walter Schulz]

garry23, are we revisiting "Santa's wish list"?
As I mentioned there: The whole PTP thingy is missing.

POC for ML interface supporting touch functions is there. I sense a growing gap between ML's menu concept and Canon's.
We lost some lens functions with newer cams (650D/M and newer).
Advanced audio options for D5?
a1ex mentioned it somewhere: Getting rid of some FRSP restrictions.

If there is a way to get realtime preview for long-time exposures into Canon cams: Nice!
For some unattended operations (yes, it's a niche) it would be great to have hardware watchdog and a alarm clock able to wake cam from sleep without external devices. Don't even know if such a function is there within Canon cams.

Nope, I do not believe we reached Ultima Thule. See sticky post "Features that are NOT possible ...". Got revised once ...

[Levas]

for my (growing) family.

Congratulations Alex!

So gonna lose some sleep in the future 

And about ML, just keep doing what you think is right, worked out pretty well over the last years.

[calle2010]

What a serious turn this April 1st thread made.

I'd like to add some Twitter wisdon from Kent Beck:

If there’s one lesson I would like the next generation of developers to learn, it is to spend less time doing hard things and more time making hard things easy. Customers benefit from the former. Customers and peers and we ourselves benefit from the latter.

I think Alex' work is fantastic especially from this point of view. The work on Qemu, documents like HACKING.RST or the M2 walkthrough are more important than another feature for ML on a specific camera.

[IDA_ML]

Ouch, that hurts and it is mostly fake news. In the past couple of years since I got a 7D I've played around with doing a firmware update which was no small task because at the time I had to use the dot-dumper. I tried a few times to get 10/12bit working and after a few false starts I finally got it working thanks to a1ex's help. Note that this happened just one month ago--not 2.5 years!

You definitely misunderstood my words.  You and I have discussed the 7D issue for quite some time and it was you saying that you are receiving very little help from the other developers and have reached the limits of your abilities.  The fact that thanks to A1ex's help you implemented these remarkable improvements into the 7D a month ago is the best proof of my statement how, with some helpful hints on his part in the right direction, talented people like you can make the miracle happen.  With 2.5 years I meant the period 2016-2018 when, despite your enormous efforts, there were lots of problems with the 7D.  This is the period when 7D shooters, waiting with great anxiety for the fixes from a month ago, simply got frustrated, gave up on their 7Ds and transferred to other camera brands.  I witnessed that and it was really painful to watch them going.  If these same fixes would have been made in 2016, the 7D would have been the 2-nd best ML camera and all these highly motivated enthusiastic people would still be proud ML users.

==============================

The title of this thread says: "Share your thoughts" and in my understanding, this is what A1ex expects us to do.  And this is exactly what I am trying to do too.   I am not blaming anybody for anything and am not complaining either.  On the contrary, I have stated so many times on this forum how much I appreciate the work of our developers and all other users involved in this remarkable international effort called Magic Lantern.  In this thread, I am just sharing my thoughts on how, again in my opinion, we can make ML work better and more efficient, convince the users of the usability of its remarkable features and keep their interest alive.  I am sorry to see that people whom I respect very much misunderstand my words and find them not positive enough.  This is probably attributed to the fact that English is not my mother tongue and I am unable to express myself in the best possible way but I will keep trying.

Anyway, I am happy to see that this important discussion is happening and evolving.  There is an old saying:

"The truth is born in discussion."

If all participants in this thread share their honest and constructive thoughts, we will finally find the truth that will help A1ex chose the best approach for himself. 

[KelvinK]

This is the period when 7D shooters, waiting with great anxiety for the fixes from a month ago, simply got frustrated, gave up on their 7Ds and transfered to other camera brands. 

I barely can understand (with overall camera prices drop and different alternatives) keep shooting for real with 900g outdated APC-S camera with moire in 2019... Literally, I don't understand whom are you talking about. It's nice (maybe) to remove some dust from your first 10 years old camera and click shutter few times here and there, but more?

[12georgiadis]

7D could have been the 2nd greatest ML camera after 5D3 but this didn't happen... I was waiting for it for a long time. Let's move forward. I'm going to buy an M50 and make some tests on it. Dual Pixel, 4k, evf... RAW soon... great program !
For MLV app, I'm ready to help financially for a Cuda GPU option. And give a lot of feedback and advices. Fast Cinema DNG is good but it's too different from ML. I helped them but I think it would be faster and better to implement CUDA in MLV app. It's good to get a fast Cinema workflow, offline/online conformation and all things that is required to use ML in the best conditions when working with hundreds of footage on a laptop ;-)

[IDA_ML]

I barely can understand (with overall camera prices drop and different alternatives) keep shooting for real with 900g outdated APC-S camera with moire in 2019... Literally, I don't understand whom are you talking about. It's nice (maybe) to remove some dust from your first 10 years old camera and click shutter few times here and there, but more?

Yes, it is an old camera but Magic Lantern really makes it shine.  The advantages of the 7D have been discussed many times on this forum and since we are running off topic, I will mention just two of them that I consider the most important ones:

1)  With a card write speed of 90 MB/s the 7D is the 2-nd fastest ML-capable camera after the 5D3.  This reveals a great potential for high-resolution RAW video recording in the cropped modes;

2) The 7D has the best cinematically looking video among all ML-capable cameras, even better than the 5D3.  The rich vivid colors, smooth tone transitions and this enormous dynamic range it provides, cannot be compared with any other video capable Canon DSLR.  What I like most is the fact that 7D MLV files have this huge processing headroom and are very easy and a lot of fun to process with both: MLVApp and DaVinci Resolve.   A properly exposed 7D video never falls apart even if you push the sliders to extremes. 

A while ago, Dfort sent to me this:

https://shotonwhat.com/cameras/canon-eos-7d-camera

Looking through the different cameras, I was quite surprised to see that the 7D was used more often than any other Canon DSLR camera in shooting serious film productions.  Film makers know what they are doing!

Finally, I have several friends filming with the 7D and ML for living - RAW videography of nature and wild life.  When talking to them about the 7D, they say that there is nothing in this world that can make them use another camera for what they are doing.  Certainly, they have very good reasons to state that.  Soon after Danne implemented high-resolution crop recording into the 100D, some of them transferred to the 100D for one single reason - it is smaller, lighter and easier to carry around in areas that are hard to access.  But they all say that, despite the lower resolution, video quality from the 7D is much better.