Magic Lantern releases new HDR Firmware for Canon T2i/550D/60D etc

** NOTE: USE WITH CAUTION. Please add to your cameras at your own risk! Hope you have a great Christmas!**

Philip: I have installed the 5DmkII version on my 5DmkII and will be trying it out myself…


Magic Lantern Warning: (As of September 2009 the software has been downloaded over two thousand times and there have been no reports of damage to the cameras. While this is no guarantee of absolute safety, the stable releases have been tested by beta testers. Most of the risk is to the developers’ cameras while testing new features and probing new portions of Canon’s firmware. By the time the software moves from development to beta testing it has been installed hundreds or thousands of times).

Introduction

By Preston Kanak

Recently, the guys at Magic Lantern released a new update to their software that brings HDR mode to Canon DSLR’s — just in time for Christmas! I have yet to play with the new software but will be over the next couple days and will be posting a video with my results. After reading a few different approaches to the post-production, it seems there is still the workflow that needs to be fine-tuned.

The Firmware – Breakdown & Analysis

Firmware Specs – (Release Notes)
– Currently running on 550D/T2i, 60D, 600D/T3i, 500D/T1i and 50D. Very
soon, it will be working on 5D Mark II and 1100D.
– User interface is evolving to be more self-explanatory and intuitive.
– The most often used essentials have been polished and refined.
– External monitors are working very well, thanks to SmallHD.
– Dedicated playback features HDR preview and exposure analysis
– Ability to select frame rates

Battery life:

– ML only draws 5% more power under heavy load.

What is it?

Magic Lantern is an enhancement atop of Canon’s firmware that frees your Canon DSLR, allowing you to use many useful features. Initially developed for filmmakers, it now has functionality for both photo and video enthusiasts, including manual audio, zebras, focus assist tools, bracketing, motion detection and much more.

With the new version, you are able to achieve an HDR-like image. Although not true HDR, this function adds a new technique when shooting with Canon DSLR’s.

How it Works

You will get two streams at half frame rate. For example, if you record 25p, you will get two 12.5fps streams. Interpolate them to 25fps and use your favorite HDR software to blend the frames.

Magic Lantern HDR Video X-MAS teaser

Bart@RedKiteMedia has created a video utilizing the new firmware. I have included it below…

Vimeo Video Description

Magic Lantern is a custom firmware add-on for Canon DSLR cameras. It is not a hack, or a modified firmware, but it runs alongside Canon’s own firmware, booting from the card every time you turn the camera on. The only modification to the original firmware is the ability to boot software from the card.

How to Install/Use

To download the firmware, please visit magiclantern.wikia.com/wiki/Unified.

Once you have downloaded the file, simply copy it to your CF card and navigate to the “Update firmware”. The active firmware on your camera temporarily shuts down as the new firmware is loaded. **One thing to note is that the firmware is loaded into the RAM which means that when the CF Card is ejected, the normal firmware that is located on your camera will reload.

How to Process the Footage — Work In Progress

If you would like to play with some of the footage, I have included the video below from a1ex.

My Thoughts

Because the workflow is still a work in progress, I have included a few different examples of how you can prepare the footage. Comments are encouraged if you have found a better way to deal with the footage!

David Fulde’s Workflow and Comments

I recently chatted with David Fulde about his process and comments on the firmware. I have included both below.

Magic Lantern HDR Video Workflow

David Fulde’s – Workflow Option 1

So, there are a couple of ways to try and shoot HDR, and a fair amount of ways to process it. Right now, I’ve been doing research into two ways.

The first method, was the first way that I thought of while playing with the footage. You need After Effects and Premiere.

1. Open Premiere. Bring your footage inside. Replace with After Effects composition (this will become important soon)

2. You should now be in AE. Take your footage, duplicate it.

3. On the top layer, right click, Time stretch – 50. This should get rid of the flicker and leave you with a 12 (Or 12.5) FPS clip which is completely your dark exposure. Time stretch is used instead of a speed ramp/time warp effect because Time Stretch at 50% simply hides every other frame; rather than trying to do some interpolation and mixing the frames together.

4. Next, go to the bottom layer; and delete the first frame. Time stretch – 50 again. This will give you a 12 (or 12.5)FPS layer of your bright exposure.

5. Next, you are going to put a Luma Key onto the first layer, key out brighter (or key out dissimilar) and play with the settings there.

6. When done, go into premiere, and create a new sequence.

7. Drop your first sequence (With the clips from AE) into this second sequence. Time: 200% this will bring you up to a 24FPS video with what you had before. this does NOT interpolate the footage, rather it duplicates every frame to give you the proper length of time.

8. Now, the reason we are going to Premiere and nesting the sequence; is because if we simply do a time warp effect without nesting the sequence (Or inside of After Effects) those frames that were hidden with the time stretch will come back and the flickering will return. So by nesting the sequence we prevent the hidden frames from returning. The downside to this, is that there are a lot of motion artifacts, as well as you are dealing with a sequence in a sequence referencing an After Effects composition; And it is SLOW. (Though you could just render out from After effects, then slow down the footage; but that also takes a lot of time.)

David Fulde’s – Workflow Option 1: Programs needed – After Effects

1. In After Effects, bring your footage into a timeline. Duplicate it until you have four copies of the footage.

2. After duplicating it, take the top two layers, and pre-compose them. This will be your highlights, name the pre-comp as such. Repeat this pre-composition with the other two clips. these will be your lowlights/shadows.

3. Go into your Highlights composition – take your top clip and set the blending mode to MULTIPLY. Move this top layer one frame forward. Your shadows will be highly crushed, and this will result in a very high contrast image this is what we want.

4. Now, go into your low-lights/shadows pre-comp and select the top clip. Change the blending mode of this to SCREEN, and move it one frame forward. This will do the exact opposite of what we did in the highlights comp. Instead of making a high-contrast image with the highlights preserved. We made a high-contrast image with the shadows preserved.

5. Next you can do the luma key as talked about above.

Note: This is great, as you do not need to do any re-timing. Each frame is repeated, but it plays surprisingly smooth! Harsh motion can still create some major motion artifacts though.

For shooting, you can choose the 720P option (With more Moire, and less resolution) and shoot at 60P (Or 50P) and then go through the process of the first method; without the need to re-time (Ending up with a 30P or 25P clip respectively) this works phenomenally well and, technically, still has artifacting but it is pretty negligible (And you can just add some motion blur to one of the clips to help mask it).

This newest version of the Magic Lantern hack, not only brings HDR, but it also brings FPS override. This means that you can choose your frame rate with frame-by-frame precision. this means that you can shoot at 720P – 48FPS for those of us in North America which, when processed using the first method above, gives us a nice, even, 24P timeline.

David Fulde’s Thoughts

I am currently loving this. The novelty factor is still here, but it’s great fun trying to find the best workflows with the least amount of tradeoffs (It’s getting to the point where it seems to be, to get 24 or 25P as an output you have to choose between less resolution, more moire, and less artifacting; or more resolution with less moire but more artifacting).

The firmware seems to have a major issue with motion; the faster the motion you have; the more artifacting (ESPECIALLY when the subject is question is moving where shadows and highlights converge)

Will this be useful for absolutely everything? No. But I can see it being used a lot for cinematography pieces. For example; You would never shoot Mobius with this; but you might shoot something like Skywalker ranch or maybe a slow paced drama. Overall; It’s fun to play with, but to be used in any sort of narrative short would need to be a special case; The workflow would essentially turn every take of every shot into an effects shot.

Ginger HDR Workflow & Examples

Magic Lantern HDR Test with Ginger HDR from 19lights.

For more information on this workflow video, please click here.

Magic Lantern HDR Video TUTORIAL

By Mr.W

Canon t2i HDR Comparison Video

By Mr.W

My Final Thoughts

I will be trying out this firmware over the next couple days so cannot comment about the firmware yet. Do I plan on integrating it into my workflow at this point? Probably not — but that may change once I have played with it. However, it is extremely exciting to see what the ML guys have been up to!

15 comments

      1. It seems to be, from what I`ve heard, that it is a combination of the dual CPU`s being hard to crack and, more importantly, the 7D has a limited number of times that you can update the firmware. So after so many tries of getting ML to work (Each time you are updating the firmware) it will lock and become impossible to upgrade again (Even with Canon`s official firmware) so it`s been VERY difficult top port (If it`s even possible)

    1. That would be great, however, you can actually shoot at 48FPS using the FPS hack on this version of Magic Lantern, and then you can conform it to 24P and perform the first method I outlined in the video (Time Stretch – 50 etc.) for a true 24P video. That way you don`t have to go from 25P to 24P (Possibly introducing some artifacts)

  1. Hello there everyone!

    Phillip you are going to love this new firmware. However, the workflows mentioned above and elsewhere online are quiet disappointing and don’t produce the natural results I was looking for. I come from a post processing background, and I have created a workflow I think people will like to try using After Effects. I think some might be interested so I’ll love to share what the differences are. I’ll post a link to a tutorial asap when done.

  2. Thanks Preston for this blog.

    I think post workflow needs a specialist tool. The workflow that team Magic Lantern provided can improve a lot. Now it separates even and odd streams and interpolates them individually. But why should one guess the movement in between the even frames when all the movement information is in the odd frame. The tool must recreate all movement from the odd frame and take exposure from the even frames.
    I’m currently working with the developer of interframe to implement such a feature. And from there the next step is a HDR tools that takes video instead of stills.
    http://www.spirton.com/interframe/

    As stated in the ML Wiki there is a post-processing tool under development for content created with the help of Magic Lantern: HDR video, timelapse, slow motion, low-light shooting, focus stacks and so on. It will have a nice graphical user interface for file handling, and available settings. And cross platform where possible.

  3. Hi….

    1) I cant use Safari to Log in onto this blog. Says wrong calculation all the time… 14+3 in my book is 17… But Safari returns “Wrong Calculation”…
    Works in Firefox which I hate… 😉

    To this blog…

    Read about HDR in ML and donated as usual to the developers…

    The HDR of ML is flawed 100% unless one shoots a film with absolutely NO MOTION…

    As soon as motion appears, per definition, there will have to be flaws as in Ghosting. Because ML and the canon does not shoot the ‘happening’ simultaneously but one frame apart. That WILL create ghosting and blur-artifacts and will end up looking worse than an over/under exposed image.

    I have been creating HDR footage since years in post. And I find the best practice is to point the camera at the lightest source in the composition to be shot. THen contemplate what is going to be white in post anyway and set my exposure accordingly. The main object WILL be underexposed at first. But in post I’ll build two mattes (Takes seconds) one for high-levels and one for low. Then I’ll adjust brightness. Thats it. Yields perfect results all the time.

    Photographers has been using this technique since God knows how long and it works a treat…

    I do a lot of photography and have been doing this since many years and even though HDR has been a ‘feature’ since many years I dont use it. Doesn’t look natural… Deliberately underexposing when shooting is yielding much more natural results and requires a little MORE knowledge BUT MUCH LESS workflow and is thus MUCH faster CPU-Cycle wise…

    Just a thought….

Leave a Reply