Messages

Where can we find the latest info on MLV? Any compiled, updated list of features? I remember there being a thread for it but I can't seem to find anything now...

Okay that all makes sense. So if I take my .raw files and use cdng to make them 16-bit linear and load them into resolve and use BMD Film color space, do I need to worry about any of this? What if I do the same thing but with 10-bit log then? From the sounds of it 10-bit log will perform the same as 16-bit linear. So is there any reason to put my footage into 16-bit linear instead of 10-bit log? Sounds like 10-bit log is the way to go.

no problem with the 5d2, i remember back in june or july this was a problem.
But it had to do with the early version of  raw2dng.
Have you tried just to use raw2dng without RAWanizer or raw2cdng ?
raw2cdng has a option to output 10bit Log if you didn't know

What's the difference between.. say.. 16 bit linear and 10bit log? Trying to find an answer online but it's rather confusing.. so.. simplified for our purposes?

I've looked at pocket camera on my next purchase. But have they updated so it has lossless Deng support? If not sorry I won't but it other codec is terrible I can't go hack from raw. Raw is just clean. I'll just buy another 7d

They're beta testing raw for the pocket right now, and it should be released very soon. The other codec is still very good though, pretty close to raw.

Don't wanna start an argument.. But I am pretty sure that if I put a 16 mm on a Pocket and the same on a 5D, I will get much wider field of view on the 5D. e.g. better depth of field.

No no, you are correct. But what I mean is that a 16mm on a c-mount/ super 16mm lens is going to be designed to look like a 50mm on full frame. It's also around the same same price as an equal quality 50mm (obviously this depends greatly, but in general that's what I've found). So the normal focal lengths anyone needs is easy to get on a super 16 sized sensor, so long as you buy the lenses designed for that size sensor. The only thing is that it gets pretty pricey when trying to go ultra wide at that size... but it can be expensive to get ultra wide in full frame as well so I mean... not a big deal. Also, cinema has never used a "sensor" the size of the 5D, e.g. photography fullframe. Super35 film is about the size of the aps-c sensor, so anything bigger than that depth of field wise is actually a little excessive in my point of view. Also, depth of field relies so much more on the f-stop than the sensor size.

All I really mean by this is that sensor size isn't really a big deal at all, what matters is understanding it and getting lenses designed for the sensor. If I had no lenses and was deciding between cameras I wouldn't look at sensor size at all, I would just pick which one wins in the other specs and get lenses that best fit the sensor.

p.s. No need for there to be a fight, just intellectual discussion.

I'm comparing color accuracy to the BMCC. It's a known fact that the colors are going to be nearly identical on the pocket, and given what we've seen from prores, that's how it is. Also, I've checked out the DNGs from Brawley, they're not bad, a little soft though. Hard to judge it just off of 5 stills though, no idea what the lenses were and all the other factors.

I didn't know that about the M, pretty cool! Moire/aliasing is definitely something I forgot about with the 5D. What do you guys think about the mosaic filter thing though? Looks like it gives pretty decent results. Most likely still sticking with the 5D though because of the low light.

Well the picture style is applied regardless, you just have to install the picture styles you want to use and then set them before you shoot. The only difference with raw is that the picture style is metadata (I'm not sure how much is actually saved though). The best way to get the "flat" look (e.g. the best starting point) is to bring the footage into resolve and change the color profile to BMD Film. If you aren't sure how to do that look it up on youtube, there are a TON of workflow videos that should help.

[Color science]

Hey guys,

I'm looking for what you guys think about this topic. I got sick of waiting for my pocket camera so I looked into the raw function not-so-long ago on my 5D. Much more stable than I expected, but I still haven't been able to try out the full power of it (only have a 133x card, just ordered a 1000x).

With how Blackmagic was going... it was obvious which I was going with. But something just threw a wrench in my plans... I just received news that my pocket camera will be shipping to me either the end of this week or beginning of next week. I am completely torn between which one to go with. For the record, I only really care about the raw function of both cameras. I will be using both for short films/a feature I have planned next year. Prores is obviously sh*t tons better than h.264, but honestly, whenever I wouldn't be filming raw either or is completely fine with me, it's never anything serious.

So lets assume the pocket camera gets updated firmware and is able to now shoot raw tomorrow (won't happen, but lets pretend). Which would be better? Lets make a list (in no particular order):
Color science: Personally, I think Blackmagic is slightly better than Canon (just based on the 2.5k version vs. current 5D raw). But they are both so good this really isn't an issue for me.
Resolution: I only film my narrative stuff in around 2.35.. so being able to do 16:9 (pocket camera) isn't a problem  to me. However the pocket still does win with the 5D's slightly lower than 1080p resolution. But lets be honest, the small difference in resolution is hardly noticeable.
Battery life: 5D obviously wins this no question. Not a huge deal, but 30 minutes (ish) per battery (pocket) is a real pain in the a**.
Overlays/Scopes/Gui: Oddly enough, again I feel the 5D wins this one. You wouldn't expect the hacked camera to have better functions but honestly... there is just more in Magic Lantern than Blackmagic. I'm not too surprised though... seems like the people here at Magic Lantern are more experienced/skilled in creating firmware than Blackmagic.
Raw files: 5D adds one extra step... but that's no big deal. As far as the quality of the raw files... as far as I am aware their versatility is the same across the board, no? Obviously the only difference (seemingly) will be the compressed CDNG from the pocket. Assuming Blackmagic isn't lying and it'll be lossless, then that's no problem.
Sound: 5D doesn't have sound so it loses here... but I do dual system sound anyway, and as it is, anyone who doesn't do that with the pocket is looking to get not so hot sound. So another "meh" point.
Lenses/Crop factor: Honestly, I don't see a big deal with crop factor, so long as you're aware of it and get the correct lenses. Tons of great c-mounts out there. So I think this comes down to purely personal taste, if even that. So it's a tie. 5D personally wins for me though since I already have a lens collection.
Stability: I can't say I trust the 5D more than Blackmagic when it comes to stability when shooting raw. It's just an innate uneasiness with using a hack. However, I am completely confidant with Magic Lantern, and rationally I know there is no risk (aside from running into problems while shooting, no long term risks). But even with the short term risks on set... I really still don't rationally find an issue with this. So I feel like it should be a tie, but my irrational fear gives a win to the pocket.
Light Sensitivity: 5D wins hands down. No question. Honestly, this is the main reason why I'm leaning more towards keeping my 5D and forgetting about the pocket. I understand lighting is very important, and I love doing it, but there's nothing like finding that creepy location in the middle of the woods. No way I can get lights there!


So... I hope that wasn't too long. What do you guys think? It seems like to me, price aside, these two cameras are nearly equal when it comes to what's available when shooting raw at a 2.35 aspect ratio. They both have their advantages and disadvantages, but nothing too crazy. It just feels weird to me that I would choose the hacked camera over the camera designed for the same function. What do you guys think? I'd be very interested in seeing other peoples opinions based  upon their own viewpoint/needs.

Had no idea about this, thanks for the quick reply.

No problem, there's a ton of stuff to know to use ML raw. That was one of the things I was unsure about when I first started too.

Is the file extension a .rar? If so, it's a spanning file and you need to throw it into raw2cdng with the original file it's linking to.

It's a spanning file. Throw it into a program like raw2cdng, or whatever you use, with the original .raw file and it'll work just fine.

I can't help you on the fine tuning, but what I can tell you is that the camera will create a single .raw file up until 4GB is reached. After that it creates a new file (I think it's a .rar) called a spanning file. When you go to extract the DNGs just drop all of the files into the extractor (I use raw2cdng) and it'll do everything for you.

Hope that helps!

It depends of the build you use.
Builds here: -> https://bitbucket.org/a_d_/magic-lantern/downloads may had the 1880px capability. But builds here: -> http://builds.magiclantern.fm/#/ don't.

Oh okay I see that. But why the difference? Different people working on those builds? If so, why would they have a difference in the max pixels?

I just installed the latest build, upgraded from the basic alpha version. I'm very confused though, why is the max horizontal pixel length now 1872? It was 1880 on the last build. I understood why it was 1880 and not 1920 (canon just upscales to full HD from 1880). Either way, why did we lose those 8 pixels?

I had some really really weird experiences when exporting with Resolve. If I do ANYTHING else when exporting I get frames that aren't graded, lines on the footage, and all kinds of other different problems. What I suggest is try a slightly less intensive version of DNxHD and once you hit start job (or whatever that button is) just don't touch anything until the render is done. If that works I'd then try a better version of DNxHD and see if that works, up until it's just too much for your computer to handle. I'm honestly not sure if that'll do it considering your problem, but maybe it's worth a shot? It's the best I can suggest.

Also, just for fun try exporting in something other than DNxHD and see if that works. If that works but all flavors of DNxHD don't, then you know it's a DNxHD problem, but if no other codec works, then you know it's a Resolve problem.

Because what comes out of the camera is 100, 200, 400, etc, iso. 160, 320, etc etc are applied as digital gain, aka processing done inside of the camera. This matters in h.264 because that is what is written to the file. However, the raw file is still only 100, or 200, or whatever you shot it at. When you open the raw file up in your editor of choice the digital gain is just  metadata. Basically, something that you can change at will. You can take a 200 iso raw file and make it look exactly like the 160 iso raw file. When you open the file up in ACR there is a whole lot of metadata being displayed. That isn't necessarily bad, but it's faulty to think that's what is coming off of the sensor. Open the same DNGs up in Resolve and switch the colorspace to BMD film. That will give you a flat log profile that comes right off of the sensor.

Now, I haven't been able to do my tests yet to confirm this 100%, but that is how I understand it to work. I know for a fact that's how metadata and ACR works. Now whether or not the difference in raw files between 160 iso and 200 iso is just metadata... I'm going to have to test that, which I definitely plan on doing.

However what I do know is that anything other than the native ISOs (100, 200, 400, etc) when shooting raw is processing applied after coming off the sensor and by definition raw files do not have any processing burned into the image.

For iso issue..

Read:
http://photocascadia.wordpress.com/2011/05/13/canon-eos-5d-mark-ii-iso-noise-test/

Also in there is a link to a well known tech company who also did a similar test.. results are there for you.

Yeah this is completely invalid for raw recording. This is how the camera acts with h.264.. Completely different beast with raw.

Well I just figured it would adjust the aperture to dial in the exposure.

Will doing auto ETTR as a starting point still work with a fully manual lens?

Once I get my 1000x card I'm going to do some tests. Right now I have a 133x so the resolution just isn't high enough to make a good judgment, eventhough it's technically the same. So far though I've had raw zebras work wonderfully for more. We shall see.

Hey,

For starters, I'm using the current stable alpha version of Magic Lantern (not sure of the exact build, but I can figure that out if necessary). I just started getting into this whole raw thing with my 5D II so I don't have a 1000x card yet. Only my old x133 card. Either way, I wanted to shoot some test footage and basically see what the workflow is like. So I shot everything at 960x408 (2.35 aspect ratio). No dropped frames, continuous recording. Awesome. Everything went well.

Until I opened the DNGs up in resolve and saw this: http://www.youtube.com/watch?v=G0IExrhE1Ag (I upscaled to 1080p because... why not)
I shot and cut together an incredibly short piece to really get a feel for how this camera works on a narrative project. Didn't get any sound though (didn't feel like it/I've done sync sound before so I'm not worried about that).
If you didn't notice, there are 100s of pink dots (especially in the first two shots) and weird colored borders on all shots.

I was going to upload some DNGs and ask what was going on... but then I decided to take a look in photoshop. To my amazement each clip is completely clean -- no glitches in sight. What in the world could be making resolve do this? I extracted the .raw files using raw2cdng1.3.0 and used bitdepth 16 | linear. I don't think that's it since the DNGs are clean as anything with photoshop (actually camera raw 8.2, but whatever) but I figured I'd still mention it.

So yeah, any idea what could be going on? If I missed any vital information just let me know. Also, if DNGs would still be useful I can upload them. What would be a good site to use though? Not up to date on what's best to use for you kids and your file sharing.

Thanks for any insight,
Robert Roth

Hold on I'm confused. I know very well about ETTR. Zebras really is the best way to do that. I'm using the newest complete "alpha" version (I can get the exact build if you want then). Not any of the nightly builds. When I go into the zebras menu I have an option for just regular zebras and then RAW zebras. Are you saying that the RAW zebras is based on the display? From my experiences (I would have to do tests) I would have to say that is not true at all. I have gotten completely different results with both zebras. If someone can confirm that RAW zebras is still just based off of the display that would be very helpful.

Also... if that is the case, labeling it is RAW zebras and basing it off of the display is extremely misleading.

Could you give a link to the newest raw2dng? I can't seem to find it either.

I seem to be having the same problem here. Why shouldn't we use raw zebras?