Comparison - SanDisk Extreme Pro 64 GB 160MB/s vs Lexar Professional 1000x 64 GB

[Protoltype]

Hey guys and maybe 0,2% girls

I did put a pdf document together for all of you who are interested for the following Compact Flash cards with best results in Magic Lantern CONTINUOS RAW Recording:

• SanDisk Extreme Pro 64 GB 160MB/s
• Lexar Professional 1000x 64 GB

And it's of course for free

In the link below you can download the pdf document.
First I did many benchmarks with Magic Lantern's CF Benchmarking Tool.
After that I tested both cards for continuos RAW Recording at different Resolutions and FPS which were maximum possible (at this time of Magic Lantern's possibilities).
What I also decided to do is comparing the two different RAW RECORDING MODULES (raw_rec & mlv_raw) at their top speeds.
Check out that document before posting too much questions

Here we go (first click on the link, then right-click and save the pdf document to your desktop):
http://www.directupload.net/file/d/3502/kuyqmw8d_pdf.htm


If there are any suggestions to make that document more readable or include other things, don't hesitate to leave a comment for that.
Any other questions besides which are answered in the document itself are welcome

[AlexRo]

The link doesn't work 

[heavygrafix]

link isn´t working. 

i am very interested.
We tried the 256Gb Sandisk 160MB/s but it isn´t faster in reality then 64 GB 1000x and 128GB KB 1050x.
The 64Gb Sandisk is here also and waiting for some tests. hopefully able to achieve 2,5k continuous without DualCard ;-)

[Steven]

It would be great if you could try to include the just announced 1066x cards from Lexar!

Also very interesting would be to find out, if the new sandisk and lexar cards higher capacities (>64GB) now perform at least as good as the 64GB cards or if theiy are even better. Because most 1000x cards have their peak performance at 32 and 64gb. Above the performance usually became worse.

[erek]

so with the SanDisk Extreme Pro 64 GB 160MB/s it is possible to continuously record 3.5K til the CF card fills up?

[trial/error]

One thing I was looking for in your SanDisk 160MB/s CF benchmark was the recording time one can get at various 60p resolutions allowing post-stretching to 16:9 aspect ratio. These resolutions should be something like: 1920x672 (16:9), 1600x560 (16:9), 1472x516 (16:9)...

I'm not sure if these resolutions are accurate and I believe that there is also another 16:9 resolution available between 1920x672 and 1600x560. Could you please test most of these (especially the highest 3 or 4) at 60p and post the recording times you get with this CF?

I want to stress once more that I am only interested in resolutions stretchable to 16:9 aspect ratio shot at 60p (not 1:2.35 or 3:2). Thanks...

[Protoltype]

The link doesn't work 

Link is working for me.
Guys is the link working now or really not.
Wait I will upload to another site.

Here is a new link:
http://www.docdroid.net/8813/cf-sandisklexar-ml-comparison.pdf.html

[Protoltype]

It would be great if you could try to include the just announced 1066x cards from Lexar!

Also very interesting would be to find out, if the new sandisk and lexar cards higher capacities (>64GB) now perform at least as good as the 64GB cards or if theiy are even better. Because most 1000x cards have their peak performance at 32 and 64gb. Above the performance usually became worse.

Yeah you're right. Unfortunately my budget allowed currently only the 64GB ones. Higher capacities were far too expensive.
But I have to say that in general I like to work with lower capacity cards and more of them instead of just one 256GB one for example. Because if anytime a 256GB CF card fails all of your footage will be gone. But instead if you have got 4x 64GB cards and just one card fails then u got 3x64GB footage. Better than nothing

[Protoltype]

so with the SanDisk Extreme Pro 64 GB 160MB/s it is possible to continuously record 3.5K til the CF card fills up?

Yeah you're right I did achieve this resolution in crop mode RAW:
3584 x 716 (5:1!) @ 24fps

But the important result was that I was only too achieve it with the raw_rec module (not the newer mlv_raw module)

[Protoltype]

One thing I was looking for in your SanDisk 160MB/s CF benchmark was the recording time one can get at various 60p resolutions allowing post-stretching to 16:9 aspect ratio. These resolutions should be something like: 1920x672 (16:9), 1600x560 (16:9), 1472x516 (16:9)...

I'm not sure if these resolutions are accurate and I believe that there is also another 16:9 resolution available between 1920x672 and 1600x560. Could you please test most of these (especially the highest 3 or 4) at 60p and post the recording times you get with this CF?

I want to stress once more that I am only interested in resolutions stretchable to 16:9 aspect ratio shot at 60p (not 1:2.35 or 3:2). Thanks...

Allright I will do that tomorrow with the SanDisk 160 MB/s CF 64GB.
I did send the Lexar one back because for me SanDisk won this game.
But for now I can say that I did try 1920x672 (16:9) @ 60fps, it did not work for me either with the Lexar or with the SanDisk CF.

I decided to take the SanDisk one also because I never ever had any problems or fails with SanDisk CFs before.
All my CFs are from SanDisk. With this new one now I have 5 CFs from SanDisk (2x 16GB 60MB/s; 2x 32GB 60MB/s; 1x 64GB 160MB/s)
Whereas I read about Lexar many people complaining that their Lexar CF 1000x 64GB failed after a few month or a year later.
I never had a Lexar so I didn't want risk anything as I do shoot weddings. I would go completely blank if this would happen.
It depends. But we will see how long my SanDisk can handle RAW shooting


Guys unfortunately I did not hear about the Lexar 1066x ones. But they seem to advertise the same speed and features as the SanDisk 160MB/s ones.
As soon I get my hands on new stuff I will make new benchmarks and post these in this forum to you and others who are thinking of buying the best for the price.
I'm glad that I could help at least some people

[Danne]

I get continous with my transcend 1000x 128gb in 3584 x 716 (5:1!) @ 24fps aswell. Wouldn,t film anything like that but still, it works.
//D

[erek]

i only payed attention to the 3584x   darn!  so 2.5K (2432 x 1366) work's continuously?

[Protoltype]

One thing I was looking for in your SanDisk 160MB/s CF benchmark was the recording time one can get at various 60p resolutions allowing post-stretching to 16:9 aspect ratio. These resolutions should be something like: 1920x672 (16:9), 1600x560 (16:9), 1472x516 (16:9)...

I'm not sure if these resolutions are accurate and I believe that there is also another 16:9 resolution available between 1920x672 and 1600x560. Could you please test most of these (especially the highest 3 or 4) at 60p and post the recording times you get with this CF?

I want to stress once more that I am only interested in resolutions stretchable to 16:9 aspect ratio shot at 60p (not 1:2.35 or 3:2). Thanks...

So here are my new results with the SanDisk 160MB/s 64GB CF + ML Build Aug22 (raw_rec module):

Continuos recording with maximum resolutions @ 60FPS & Aspect Ratio 16:9 (Global Draw = Off, Small Hacks = ON)

• 1600x560 (Stretch in post by 1.61x to get 1600x900) BUT ONLY WITH PREVIEW MODE on "HaCKeD" (so no live-view while recording)
• 1472x516 (Stretch in post by 1.61x to get 1472x828) PREVIEW MODE "Canon"

Of course less than 1472x516 will also be successful
Hope this helped.

[Protoltype]

i only payed attention to the 3584x   darn!  so 2.5K (2432 x 1366) work's continuously?

So 2560 x 1024 (2.50:1)@ 24fps is working sometimes continuosly but sometimes just one minute.
I think it depends on how much color information is send through the camera's sensor...
But that is just my estimation.
The developers can maybe give a more detailed explanation of that.

If you want to shoot for production with 2560 x 1024 (2.50:1) @ 24fps shots that are longer than one minute I would not recommend it.
But anything below that is fine

I think a common good movie will not have shots longer than one minute except it is a prepared steadycam shot without any cuts.
But it depends as always

[trial/error]

So here are my new results with the SanDisk 160MB/s 64GB CF + ML Build Aug22 (raw_rec module):

Continuos recording with maximum resolutions @ 60FPS & Aspect Ratio 16:9 (Global Draw = Off, Small Hacks = ON)

• 1600x560 (Stretch in post by 1.61x to get 1600x900) BUT ONLY WITH PREVIEW MODE on "HaCKeD" (so no live-view while recording)
• 1472x516 (Stretch in post by 1.61x to get 1472x828) PREVIEW MODE "Canon"

Of course less than 1472x516 will also be successful
Hope this helped.

Thanks, it helped. So is there one more 16:9 resolution available between 1600x560 and 1920x672? If so, how long can you record 60p and 50p with this CF at it? I know it's not going to be continuous, but how many seconds would I typically get? Thank you again!

[Protoltype]

Thanks, it helped. So is there one more 16:9 resolution available between 1600x560 and 1920x672? If so, how long can you record 60p and 50p with this CF at it? I know it's not going to be continuous, but how many seconds would I typically get? Thank you again!

Yeah there are 2 more resolutions betweeen these:  1728x606 and 1856x650

1600x560 @ 60fps = 13 seconds
1728x606 @ 60fps = 5 seconds
1856x650 @ 60fps = 3 seconds

1600x560 @ 50fps = continuos !
1728x606 @ 50fps = continuos !
1856x650 @ 50fps = 9 seconds
1920x672 @ 50fps = 5-6 seconds

All tests with:
SanDisk 160MB/s 64GB CF
ML Build Aug22 (raw_rec module)
Global Draw = Off
Small Hacks = ON
Preview = Canon


I think the bottleneck here is not the card itself.
It is the Canon 5D Mark3 or Magic Lantern.
The Canon 5D Mark3 is probably overloaded with processing when it records with 60fps so there is not enough processing power left to write the RAW data with ML fast enough.
If I am wrong, please someone correct me.

But why are you so hungry to record with 60fps raw.
I think even 50fps is not worth it shooting raw because the quality is not so beautiful like the 24fps raw 1920x1080.
The vertical resolution is squashed hardly so you loose information which you have to stretch in post....

For slow motion I would better shoot with 24 or 25 fps raw continuosly and slow the footage down in post with the plugin twixtor or the foundry's kronos.
Glad to help you out

[Danne]

Have you seen the quality of unsqueezed 50fps? It is beautiful. I, d revommend using it any day of the week.

[trial/error]

Yeah there are 2 more resolutions betweeen these:  1728x606 and 1856x650

1600x560 @ 60fps = 13 seconds
1728x606 @ 60fps = 5 seconds
1856x650 @ 60fps = 3 seconds

1600x560 @ 50fps = continuos !
1728x606 @ 50fps = continuos !
1856x650 @ 50fps = 9 seconds
1920x672 @ 50fps = 5-6 seconds

All tests with:
SanDisk 160MB/s 64GB CF
ML Build Aug22 (raw_rec module)
Global Draw = Off
Small Hacks = ON
Preview = Canon


I think the bottleneck here is not the card itself.
It is the Canon 5D Mark3 or Magic Lantern.
The Canon 5D Mark3 is probably overloaded with processing when it records with 60fps so there is not enough processing power left to write the RAW data with ML fast enough.
If I am wrong, please someone correct me.

But why are you so hungry to record with 60fps raw.
I think even 50fps is not worth it shooting raw because the quality is not so beautiful like the 24fps raw 1920x1080.
The vertical resolution is squashed hardly so you loose information which you have to stretch in post....

For slow motion I would better shoot with 24 or 25 fps raw continuosly and slow the footage down in post with the plugin twixtor or the foundry's kronos.
Glad to help you out

Thank you again, this kind of information was hard to find. I'm so into slow-motion because I'm hoping to use the Mark lll's raw for music video production. Creating fake frames using Twixtor is either a big "no-no" or really time consuming in its full version. You would need to mask everything, etc. Unnecessary hell. 1728x606 looks promising Kudos for doing all these benchmarks for us!

[Protoltype]

Have you seen the quality of unsqueezed 50fps? It is beautiful. I, d revommend using it any day of the week.

Yeah you are right.
I am just spoiled because I did shoot with a Red Epic recently, but it is not fair to compare both cameras because of their huge price difference 
I fell in love with it. But of course I cannot afford it  The 5D Mark 3 with ML RAW is coming reaally close to it's image quality (in non-slow mo).

[Protoltype]

Thank you again, this kind of information was hard to find. I'm so into slow-motion because I'm hoping to use the Mark lll's raw for music video production. Creating fake frames using Twixtor is either a big "no-no" or really time consuming in its full version. You would need to mask everything, etc. Unnecessary hell. 1728x606 looks promising Kudos for doing all these benchmarks for us!

Of course you are right it is always better to shoot slow mo straight away
No problem I am happy I could help.

[Danne]

@protoltype
I,d be interested in a comparison between red epic and ml raw in let,s say 50fps and 1920x606. Could be done in 5 seconds clips.

Here,s a comparison between two aspect ratios on a 5d mark 3 and a tutorial how to restretch several clips in 50fps in after effects.
Conclusion

(16:9) 1728 x 606 50fps works continuous and works fine interpolated to 1920x1080

Other fine alternative aspect ratio continuous is 1920x544(2:20:1)

[Protoltype]

@protoltype
I,d be interested in a comparison between red epic and ml raw in let,s say 50fps and 1920x606. Could be done in 5 seconds clips.

Here,s a comparison between two aspect ratios on a 5d mark 3 and a tutorial how to restretch several clips in 50fps in after effects.
Conclusion

(16:9) 1728 x 606 50fps works continuous and works fine interpolated to 1920x1080

Other fine alternative aspect ratio continuous is 1920x544(2:20:1)

Thank you I'll check that out.
As for the comparison between Red Epic and 5D Mark3 ML Raw, maybe in a few month I can get my hands on the Red Epic than I'll definately do a image quality comparison.
But not only the image quality. The post production workflow with r3d files is so easy and smooth. R3D is also RAW but here you can go in and define the compression strength.
The files are supported by almost any NLE. It is GPU accelerated and the rendering is muuuuuuch quicker than 5D Mk3 RAW DNG Files. Although the Red shoots with higher resolutions.
But I think in maybe 1 year the ML RAW DnG files will be supported aswell and render GPU accelerated. This would be a great feature !

[Danne]

Please do a comparison. Also check this guys video in 50fps http://www.magiclantern.fm/forum/index.php?topic=9534.0

[WorzelG]

Any comparisons yet with the 1066x card??

[ibrahim]

This may be a silly question but in order to shoot 1080p continuously at 24fps you used to cards, an empty SanDisk Extreme Pro 64 GB 160MB/s as well as INX 1GB SD card with ML installed on it?

If so, why use two cards instead of using ML on the CF card alone?