Magic Lantern... CONTINUOUS raw recording @ 24fps on 5D3

[rpt]

Here's some footage I edited together from last weekend in Central Oregon:

https://vimeo.com/66866250

All but the last three shots are with the Magic Lantern firmware....it's a bit tedious in post but WORTH IT!

Great video! Your intro reminded me of some of the other amazing bucolic footage I've seen from other ML RAW early adopters.

I was able to get RAW working on my 5D3. Tomorrow I'll shoot some footage and string it together. But alas, I don't think it will include misty mountains and green pastures. ;-) Lol... It will probably be more like bumper-to-bumper traffic, construction work, and people yelling at each other. Ah... city life.

So you guys went back to 1.1.3 or never upgraded to 1.2.1? I shall be watching this channel for your videos and comments

I never upgraded because I heard it breaks Magic Lantern. I had never used Magic Lantern prior to the RAW video thing, I'm enjoying all the other, many features as well!

I've heard it's pretty easy to downgrade back to 1.1.3 though!

Yes, I know. However, my f8 focus will go away and my flash focus will be even lazier than me (or should have I said "lazier than I").

I'll wait on the sidelines for now - I have the 1.1.3 saved and so I could just make the jump...

Or else I shall watch the videos you people put up and drool...

Whatever...

Have fun!

Ahh, I see, I don't have a lens setups fancy enough that I need f/8 focus.

I have no doubt this will *eventually* work with 1.2.1 though, the rate at which these guys are fixing bugs is pretty impressive. Most of my complaints about the raw2dng OSX GUI software have been fixed within days!

Actually I do not have fancy glass. If I did, I would not need the f8 focus. I have the 100-400 which is a f5.6 at the long end and when I add the 1.4x, it becomes a f8 560mm. If I had the fancy 500mm or 600mm f4 like some folks do, I would probably not upgrade... So I'll wait... And look forward to videos put up by you folks

 

[bchernicoff]

Okay, so I have been playing with this today after receiving my 32gb Lexar 1000x card. It's really amazing. The amount of color and detail is phenomenal. I don't think it's been posted here, but After Effects can work with the DNG files directly. So workflow is:
1)copy M00000*.RAW to computer
2)raw2dng M00000*.RAW
3)In AfterEffects File->Import->File
4)Choose the first DNG file only (this should cause After Effects to import as Footage)
5)Adobe Camera Raw opens and shows you the first frame. Tweak as needed...it will apply to all frames automatically.
5)Right-click the imported footage in the Project window and choose Interpret Footage->Main
6)Change the frame rate to 23.976(or whatever you shot at)
7)Create a composition with the size and frame rate as what you shot
8)Drag footage onto composition.
9)Render as whatever you like (I've been using H.264, Main profile 5.1 with 31Mbps average and 52Mbps Max)
10)...
11)Profit!

I will add though, working with the DNGs in Lightroom is still really useful though. I find it much easier to tweak and do cool stuff. I also created a PS Action to batch process images in Silver Efex Pro 2...that takes a LONG time, but was really cool.

 

[dirtcastle]

Okay, so I have been playing with this today after receiving my 32gb Lexar 1000x card. It's really amazing. The amount of color and detail is phenomenal. I don't think it's been posted here, but After Effects can work with the DNG files directly. So workflow is:
1)copy M00000*.RAW to computer
2)raw2dng M00000*.RAW
3)In AfterEffects File->Import->File
4)Choose the first DNG file only (this should cause After Effects to import as Footage)
5)Adobe Camera Raw opens and shows you the first frame. Tweak as needed...it will apply to all frames automatically.
5)Right-click the imported footage in the Project window and choose Interpret Footage->Main
6)Change the frame rate to 23.976(or whatever you shot at)
7)Create a composition with the size and frame rate as what you shot
8)Drag footage onto composition.
9)Render as whatever you like (I've been using H.264, Main profile 5.1 with 31Mbps average and 52Mbps Max)
10)...
11)Profit!

I will add though, working with the DNGs in Lightroom is still really useful though. I find it much easier to tweak and do cool stuff. I also created a PS Action to batch process images in Silver Efex Pro 2...that takes a LONG time, but was really cool.

Thanks for all the details! This is great. We have so many options now (even if they do take a bit more time). I'm definitely excited for Silver Efex. And I also prefer Lightroom, so that will probably be my workflow too.

Unfortunately, I've been having to debug CF card issues. I'm still working on how to get 1920x1080 consistently. I can get resolutions below that, but I'm not doing all this work just to settle for a compromise! ;-)

Fortunately, there's a lot of helpful discussion going on over at the ML forums. I feel like the primary issue with the RAW hack is CF card performance. I can see why Canon wouldn't bother with this: even the top CF cards barely cut it. There are only a few cards up to the task. I thought I had a super badass card (Lexar 1000x), but I'm still having to finesse it to get 1920x1080.

 

[bchernicoff]

Unfortunately, I've been having to debug CF card issues. I'm still working on how to get 1920x1080 consistently. I can get resolutions below that, but I'm not doing all this work just to settle for a compromise! ;-)

Fortunately, there's a lot of helpful discussion going on over at the ML forums. I feel like the primary issue with the RAW hack is CF card performance. I can see why Canon wouldn't bother with this: even the top CF cards barely cut it. There are only a few cards up to the task. I thought I had a super badass card (Lexar 1000x), but I'm still having to finesse it to get 1920x1080.

Yeah, now I've run into the same problem. After recording 8-10 clips at 1920x1080 on this card it's now stopping after 33 frames every time. I have a couple of thoughts on this. I know that with solid-state hard drives the fuller they become, the slower become due to the load leveling strategies that get used and the reluctance to rewrite flassh cells until all of them have seen some use. Another thought is battery power. ML started giving me a yellow battery indicator at the same time this started. I wonder if power management comes into play and some of the hardware gets underclocked.

I'll troubleshoot tomorrow, but if you have any breakthroughs please let me know.

 

[dirtcastle]

Good stuff! I appreciate the camaraderie! I've been weed-wacking this sh*t for days, so it's nice to have someone else in the same boat. I think we're having the same buffer / dropping / skipping frames problems.

From what I can glean on the ML forums, the following CF cards appear to be the best, (that said, there are differences between the 32/64/128 sizes). Most people seem to be finding 32/64GB cards as the best performers.

• Toshiba 1066x
• Lexar 1000x
• Hoodman Steel 1000X
• Transcend 1000x

My hunch about our problems is that it's probably not the card (mine is a Lexar 1000x 32GB). I've done the benchmarks with mine (92+MB/s), I've seen others get good results with the same card (and I've even had it work on occasion, though not repeatably). Even when I go to much lower resolutions, I'm still having problems. I am guessing that I can get it to work with this card. And when I get some new/different cards, I can see if that makes a difference. But it would still be nice to use cards I already own

To be honest, I'm just troubleshooting in the dark right now. But here are some of the variables I'm learning about and testing.

1. Build. I've been using May 22, but I might try May 19 (that's the one used in the http://www.cinema5d.com/news/?p=17898 tutorial).

2. Reformatting/Resetting/Reinstalling. Sometimes just wiping/resetting stuff solves problems of staleness or hidden user errors. I'm including Canon firmware, ML builds, and SD/CF cards in my wipes/resets. This is probably the most painful part of debugging, but it's like an insurance policy against stupid sh*t.

3. Resolution. I've been starting with 1920x1080, but I should probably work my way up, if only for the purpose of seeing where the threshold is. Hopefully I can use changes in the minimum resolution threshold to know if some of these variables are making a difference.

4. Canon Settings. I'm not sure whether changing the resolution/fps in the Canon native firmware makes a difference, for example between All-I and IPB. I'm just guessing that the lower the res/fps, the more likely it will work with a card.

5. HACK3D. I have no idea what this is. All I know is that some people say it helps and it turns Live View off. I'm skeptical, but little things might make a difference.

6. Warming Up. Some people have found that it takes a bit of shooting action on the camera before it will hit optimum card performance. In the words of sergiocamara93: "...if you record a couple GBs, I go all the way up to 4GB, in 1920x960 and them you switch to 1080 they are just fine after that. I've recorded several 16GB files without a problem." This sounds more like voodoo (or a corollary solution), but even if it's not a direct solution, it might indicate where the deeper problem is.

7. Global Draw. I see it in the menu, but I have no idea what it does. Some people say turning it off makes a difference; others say it doesn't. I lump this together with HACK3D in voodoo solutions.

8. Buffer settings. I don't know anything about card buffering. I looked to see if there were any options or settings for it, but I couldn't find any. There is an image buffer item in the Debug menu, but it's not editable. One of the most core forum topics on this is entitled "Card benchmark - what's the optimal write buffer size?" http://www.magiclantern.fm/forum/index.php?topic=5471.0 Strangely, I couldn't find any direct references to buffer size, only to speed and stability. So I'll assume buffer is implied by other indicators and not directly adjustable, beyond using a different card (or changing resolution/fps or other demands on the card).

9. Battery power. As you mentioned, it's worth looking into. Does the power vary as the battery drains?

10. What else could it be?? I'm sure I will have 10 more theories by tomorrow. ;-)

And of course, having listed all these individual reasons, it is probably several of them, in conjunction, that will make the difference.

BTW, does anyone know how to how to go back and forth between the Canon menus and Magic Lantern menus? When I'm in the Canon menus, I click on the trash can, but nothing happens. I'm assuming it's not possible, which would make sense from the "bootup" nature of ML.

 

[rpt]

bchernicoff, dirtcastle, have you tried low level format of the card after downloading? I thought there was a thread either on CR or ML forums about this... If that works maybe you need to add that to your workflow...

 

[dirtcastle]

bchernicoff, dirtcastle, have you tried low level format of the card after downloading? I thought there was a thread either on CR or ML forums about this... If that works maybe you need to add that to your workflow...

I was wondering about that.

Fortunately, I finally got 1920x1080 working a few minutes ago. I started all the way from scratch and used the same instructions from Cinema5D.com (except this time I actually followed them : ). http://www.cinema5d.com/news/?p=17898

Here are some of the highlights...

0. I loaded a fresh battery.

1. I reformatted all the cards and reset/cleared the Canon 1.3.1 firmware settings.

2. I used the May 19 build (linked in the tutorial) instead of the May 22 build I was using before.

3. I recorded one clip with the default resolution (1360x580). It worked.

4. When I up'd the resolution to 1920x1080, the required write speed (at the bottom of the raw video screen in the movie menu) was over 100MB/s, which is faster than my card. I noticed the fps was set to 29.9fps.

5. I went into the Canon menus and set the movie record size to 1920x1080 23.9fps All-I.

6. I turned off/on the camera and restarted ML and the raw_rec module (same as always). When I went to the movie dimensions it said it required 82.9MB/s, which is below my confirmed benchmark of around 92MB/s.

Long story, short?... I learned a bunch of interesting stuff about ML, but in the end it was starting from scratch and following directions that worked. I deviated a bit the first time because I got it to work and I didn't realize how bug-prone it is.

Now that I have a formula that worked once... I have something to fall back on (because it will probably be broken again by tomorrow afternoon : ).

 

[Drizzt321]

A comment on the write speed problem, it might be that the CF card has to have good garbage collection, because if you want to write to a NAND block that already has data, you first have to clear it, then write to it. Some of the cheaper ones might not implement that well, and it might be that the better CF card controllers can detect when a card is reformatted (rather than delete files) and mark those blocks to be cleared out in advance of need like the TRIM command that modern SATA SSDs support.

Also, I bet the reason the 32GB/64GB cards have the best performance is that they have fully populated NAND die control channels (NAND controllers like to read/write dies in parallel). They may also be using SLC memory in some cases for speed and/or longevity, while the 128GB cards might be using MLC in order to get that high which tend to be slightly slower.

I'm looking forward to what XQD can do, since it's based off of PCI-Express. It'd be pretty cool if someone were to come up with an adapter that is basically a cable (modification to camera required) that has a XQD on one side, and an external SATA/PCIe high capacity, high speed SSD. Imagine capturing the full sensor in RAW and writing it out at 24/30 fps? *drool* Heck, I wonder if with those write speeds you could do faster than 30fps...

 

[LetTheRightLensIn]

Okay, so I have been playing with this today after receiving my 32gb Lexar 1000x card. It's really amazing. The amount of color and detail is phenomenal. I don't think it's been posted here, but After Effects can work with the DNG files directly. So workflow is:
1)copy M00000*.RAW to computer
2)raw2dng M00000*.RAW
3)In AfterEffects File->Import->File
4)Choose the first DNG file only (this should cause After Effects to import as Footage)
5)Adobe Camera Raw opens and shows you the first frame. Tweak as needed...it will apply to all frames automatically.
5)Right-click the imported footage in the Project window and choose Interpret Footage->Main
6)Change the frame rate to 23.976(or whatever you shot at)
7)Create a composition with the size and frame rate as what you shot
8)Drag footage onto composition.
9)Render as whatever you like (I've been using H.264, Main profile 5.1 with 31Mbps average and 52Mbps Max)
10)...
11)Profit!

I will add though, working with the DNGs in Lightroom is still really useful though. I find it much easier to tweak and do cool stuff. I also created a PS Action to batch process images in Silver Efex Pro 2...that takes a LONG time, but was really cool.

One problem I find using AE only even when you don't need to do anything in PS is that you get stuck the file being put out in sRGB format and yet most video monitors/hdtvs/etc. use gamma 2.2 which is NOT what sRGB format uses so you get a tone curve shift which slightly increases apparent contrast/saturation/darkes deep and mid tones so then all your careful ACR edits in AE don't look quite right in Premiere Pro or when you play the files back with most software unless you calibrate your screen to sRGB TRC instead of gamma 2.2.

I guess you might be able to set the output profile under color management in AE to not use Working Space but to an .icc profile that uses sRGB/REC709 primaries mixed with gamma 2.2 though. I haven't come upon such a profile though. I think I will try to make one myself. That would solve the AE-only workflow color management issues.

(With PS workflow as I said in my other message here you just stick in a convert to Custom RGB REC709 primaries gamma 2.2 profile step as the last step in your batch action before saving/closing so you don't have to have on hand an .icc file that has the proper settings.)

 

[LetTheRightLensIn]

Okay, so I have been playing with this today after receiving my 32gb Lexar 1000x card. It's really amazing. The amount of color and detail is phenomenal. I don't think it's been posted here, but After Effects can work with the DNG files directly. So workflow is:
1)copy M00000*.RAW to computer
2)raw2dng M00000*.RAW
3)In AfterEffects File->Import->File
4)Choose the first DNG file only (this should cause After Effects to import as Footage)
5)Adobe Camera Raw opens and shows you the first frame. Tweak as needed...it will apply to all frames automatically.
5)Right-click the imported footage in the Project window and choose Interpret Footage->Main
6)Change the frame rate to 23.976(or whatever you shot at)
7)Create a composition with the size and frame rate as what you shot
8)Drag footage onto composition.
9)Render as whatever you like (I've been using H.264, Main profile 5.1 with 31Mbps average and 52Mbps Max)
10)...
11)Profit!

I will add though, working with the DNGs in Lightroom is still really useful though. I find it much easier to tweak and do cool stuff. I also created a PS Action to batch process images in Silver Efex Pro 2...that takes a LONG time, but was really cool.

You can also set AE to default to 23.976 (and 16bits) if that is how you shot most of your footage (very likely since 1920x1080x30p tends to choke out after a few seconds even with 32GB Lexar 1000x, I can only get 23.976 rates sustained). And then you can simply change step 3 to "Import-Multiple Files" and then you pick the first DNG and edit in ACR to set the template and then it pops open the file browser again and this time you hit "Import Folder" (make sure that the DNG are numerically orderded and have nothing more than a simple letter at the end of the numbers otherwise, for some reason, AE tends to get confused and not think you have a sequence and then you have to mess around forcing it to realize what is going on) and then skip straight to step 9!

I've been exporting to AVI lossless 16bit and also trying CinemarkAVI 4:4:4 HD optimized/filmscan1/2 to preserve everything best for use in Premiere Pro so I don't lose enough more on final compression out of PP and if I do end up compositing or editing colors more in PP. For files that are not Earth shattering, I think I might just keep the CInemark 4:4:4 for archival purposes to save space and delete the original RAWs (for really good stuff I feel like I'd still want to hang onto the RAWs for now and delete the intermediate footage when I feel I probably won't use it again for a while).

I'm thinking of making a deep color video, since some of my flower shots are to intense for sRGB gamut. I might make the footage in ProPhotoRGB if I can get full color management working in MPC+MadVR or I might just render them to the profile of my monitor set to native gamut (although they would've be shareable with others for the most part in this case). It would be pretty cool. Deep color video is not something the average person on the street has been able to really mess with before. But RAW video allows it. At least from Canon DSLR.

 

[dirtcastle]

A comment on the write speed problem, it might be that the CF card has to have good garbage collection, because if you want to write to a NAND block that already has data, you first have to clear it, then write to it. Some of the cheaper ones might not implement that well, and it might be that the better CF card controllers can detect when a card is reformatted (rather than delete files) and mark those blocks to be cleared out in advance of need like the TRIM command that modern SATA SSDs support.

I will test this out. Based on my experience and what you're suggesting, I think the best route will be to fill a card in sequence and then reformat after it fills up. Hell, I won't even preview clips. After I get sustained speeds, I can test and figure out what compromises the card speed.

 

[Drizzt321]

A comment on the write speed problem, it might be that the CF card has to have good garbage collection, because if you want to write to a NAND block that already has data, you first have to clear it, then write to it. Some of the cheaper ones might not implement that well, and it might be that the better CF card controllers can detect when a card is reformatted (rather than delete files) and mark those blocks to be cleared out in advance of need like the TRIM command that modern SATA SSDs support.

I will test this out. Based on my experience and what you're suggesting, I think the best route will be to fill a card in sequence and then reformat after it fills up. Hell, I won't even preview clips. After I get sustained speeds, I can test and figure out what compromises the card speed.

I'm just extrapolating based off of modern SSDs. Ah, looks like CF 6.0 introduced UDMA7, as well as TRIM command just like SSDs have. So I imagine that's what is being used during a delete/quick-format to tell the controller that it no longer needs to keep the NAND blocks permanently, and can clear them out at will. This is part of what the controller garbage collection does, and usually operates in the background. So it might be you need to leave it sitting for a short bit in the camera/reader after deleting/formatting the card to give it power and let it do it's background cleanup to keep maximum performance.

The test you're describing is part of what Ananadtech.com does during it's SSD tests, as well they now check the consistency of latency which can be important to avoid buffer overruns in a case like this, where if it has a brief spike in latency that might reduce overall throughput just enough that you start dropping a few frames.

 

[dirtcastle]

I just got a Hoodman 1000x 64GB and I'm testing it out. The first benchmark test clocked around 90Mb/s. The three after that have been over 93MB/s. That puts this card roughly in the ballpark of my Lexar 1000x 32GB (based on benchmarking). Now I'll see how it does with actual shooting.

UPDATE: Shooting 1920x1080 RAW is working great with both Hoodman and Lexar cards. I'll post footage as soon as I figure out how to shoot something decent. Video n00b.

 

[bchernicoff]

It seems like there must be something wrong with my card now. It was brand new Lexar 1000x 32gb. I recorded about 10 1920x1080 raw clips on it with no problem...usually about 3 at a time after which I would delete the files but not reformat or anything. Then it stopped working. Now it will write 116.1mb before the buffer fills. This is regardless of what resolution I try. ML shows a write speed of 6.6mb/s, so something has crippled the card. Just for the heck of it, I tried one of my older Lexar 400x 16gb cards and ML is able to write to them at 26mb/s no problem, so it's not the camera.

I've tried reformating, aligning partition to 4096, completely overwriting the card. Still, no change from 6.6mb/s. Does this sound familiar to anyone? I'm thinking that it's 3 days old at this point and I should just return it.

 

[dirtcastle]

It seems like there must be something wrong with my card now. It was brand new Lexar 1000x 32gb. I recorded about 10 1920x1080 raw clips on it with no problem...usually about 3 at a time after which I would delete the files but not reformat or anything. Then it stopped working. Now it will write 116.1mb before the buffer fills. This is regardless of what resolution I try. ML shows a write speed of 6.6mb/s, so something has crippled the card. Just for the heck of it, I tried one of my older Lexar 400x 16gb cards and ML is able to write to them at 26mb/s no problem, so it's not the camera.

I've tried reformating, aligning partition to 4096, completely overwriting the card. Still, no change from 6.6mb/s. Does this sound familiar to anyone? I'm thinking that it's 3 days old at this point and I should just return it.

The other isolation test (in addition to trying a different card) would be to try that Lexar card with a different camera. But that's probably academic at this point. Bottom line: if it doesn't work... it doesn't work. And if you can solve the problem by getting a different card... save yourself the headache.

I've purchased several cards because it's clear that the size options are a function of the card's performance. I'm returning a KomputerBay 128GB card. I'll find out how the Transcend 128GB works when I get it next week. If the Transcend 128GB doesn't work out, I'll be trying Toshiba 64GB and maybe Transcend 64GB (if I feel like pressing my luck). Considering how much we pay for these things, the ability to return it is part of the cost.

 

[AAPhotog]

Heres a test. a stupid one but nevertheless a quick recorded clip.

Shot on a 16gb Sandisk 600x CF

https://vimeo.com/66705345

 

[AAPhotog]

It seems like there must be something wrong with my card now. It was brand new Lexar 1000x 32gb. I recorded about 10 1920x1080 raw clips on it with no problem...usually about 3 at a time after which I would delete the files but not reformat or anything. Then it stopped working. Now it will write 116.1mb before the buffer fills. This is regardless of what resolution I try. ML shows a write speed of 6.6mb/s, so something has crippled the card. Just for the heck of it, I tried one of my older Lexar 400x 16gb cards and ML is able to write to them at 26mb/s no problem, so it's not the camera.

I've tried reformating, aligning partition to 4096, completely overwriting the card. Still, no change from 6.6mb/s. Does this sound familiar to anyone? I'm thinking that it's 3 days old at this point and I should just return it.

The other isolation test (in addition to trying a different card) would be to try that Lexar card with a different camera. But that's probably academic at this point. Bottom line: if it doesn't work... it doesn't work. And if you can solve the problem by getting a different card... save yourself the headache.

I've purchased several cards because it's clear that the size options are a function of the card's performance. I'm returning a KomputerBay 128GB card. I'll find out how the Transcend 128GB works when I get it next week. If the Transcend 128GB doesn't work out, I'll be trying Toshiba 64GB and maybe Transcend 64GB (if I feel like pressing my luck). Considering how much we pay for these things, the ability to return it is part of the cost.

Many of the 128GB cards are much slower than their 64gb counter parts. Why not save yourself some money and go with the 64gb Komputerbay card that are PROVEN to work with the RAW at 1080p.
I've only seen 1 or 2 cases of them not working and it was because of a dead card, not speed

 

[LetTheRightLensIn]

It seems like there must be something wrong with my card now. It was brand new Lexar 1000x 32gb. I recorded about 10 1920x1080 raw clips on it with no problem...usually about 3 at a time after which I would delete the files but not reformat or anything. Then it stopped working. Now it will write 116.1mb before the buffer fills. This is regardless of what resolution I try. ML shows a write speed of 6.6mb/s, so something has crippled the card. Just for the heck of it, I tried one of my older Lexar 400x 16gb cards and ML is able to write to them at 26mb/s no problem, so it's not the camera.

I've tried reformating, aligning partition to 4096, completely overwriting the card. Still, no change from 6.6mb/s. Does this sound familiar to anyone? I'm thinking that it's 3 days old at this point and I should just return it.

The other isolation test (in addition to trying a different card) would be to try that Lexar card with a different camera. But that's probably academic at this point. Bottom line: if it doesn't work... it doesn't work. And if you can solve the problem by getting a different card... save yourself the headache.

I've purchased several cards because it's clear that the size options are a function of the card's performance. I'm returning a KomputerBay 128GB card. I'll find out how the Transcend 128GB works when I get it next week. If the Transcend 128GB doesn't work out, I'll be trying Toshiba 64GB and maybe Transcend 64GB (if I feel like pressing my luck). Considering how much we pay for these things, the ability to return it is part of the cost.

I'd try sticking to 64GB (or 32GB) since all the almost all of the 128GB seem to run a touch slower (and some such as the Komputer Bay vastly slower).

Also I'd avoid KomputerBay at any size since I'm reading so many reports about cards failing after two weeks and talk that they use Lexar parts that got rejected.

 

[dirtcastle]

That's what I'm seeing too, that the 32GB and 64GB are the fastest. Personally, I will be avoiding KomputerBay until the dust settles. I bought a KomputerBay 128GB and it clocked around 72MB/s. I don't see any risk in ordering a KomputerBay 64GB, but be ready to return it. I've got a Transcend 128GB on order (from weeks ago, before I knew they were unreliable), but I'll be ready to ship it back same day I get it.

The known solid candidates are Toshiba, Hoodman, Lexar, and Transcend. I've got a Lexar 32GB and a Hoodman 64GB and both are solid.

 

[rpt]

One of the conversations on the ML forum they mentioned that the 128GB UDMA 7 cards were slower than the 64 GB UDMA 7 cards and they ended up dropping frames. I think they were talking about Lexar. So they were shooting on the 64GB cards...