Topics

definitely watch in 4k:

https://youtu.be/Zrm0F4ljv8U

the same technique as the others, first bessel and then 2x "sharp" superscaling.

i can't find any side effects!

watch in 4k, of course!

https://www.youtube.com/watch?v=aaMIy9UrQ_I

i needed a break from card spanning hassles so i shot this video at 2880 using only the CF card with the 5d3.

it was a pleasant surprise to see how nicely it scaled to 4k using the settings i've been using lately:

bessel algorithm for basic scaling followed by 2x "sharp" superscaling.

that's it!

it's really hard to see any aliasing anywhere, and it is sharp.
it might work well for even lower resolutions...

EDIT: another video added, shot at the same resolution and upscaled the same way, with really beautiful results.  i haven't yet found a single problem! i even pushed the exposure a bit on this one with some highlight clipping.  it was inside the veranda with occasional sunlight reflecting on the crystal vase.  no problem!

this is a really great setting for quick and easy "4k". 

https://www.youtube.com/watch?v=G9fCoU76R6A

https://youtu.be/EJP-T_w3zks

not truly golden hour because the trees are too tall, but as close as it gets before the shadows cover everything

bessel resize sharpening followed by 2x "sharp" superscaling in resolve with mlvfs using card spanning:

https://www.youtube.com/watch?v=7omYK5dTf9Y

here's a video i shot at <> 3400*1700, and there is a single frame "glitch" at about 7 or 8 minutes from mlvfs.  using separate card readers for each card definitely helped, but it is still not perfect.  this "bad frame" happened repeatedly, no matter how many times i rendered, and also across reboots of the pc.

the strange thing is that the frame is correct when scrubbing in resolve, but rendering always results in the error.

this setup is so precarious that it isn't surprising there are such issues.  it's a real drag, because it is such an amazing timesaver when it is working...

i may try different versions of dokan again to see if there's a "solution" but it may be related to so many other factors, including the peculiarities of this particular build of windows, or card reader drivers, or quantum effects due to the sun's corona, etc.

i was considering rendering the single frame as a tiff and then surgically inserting it, but even rendering as tiff resulted in the error!

https://youtu.be/We7suAl_X-w

i recently did a few experiments of shooting at lower resolutions and trying to get similar quality via tricks such as using superscaling twice (resolve), and although the results were really quite good, maybe even indistinguishable to the viewer, when i returned to shooting at 3400*1700 for this video, it was just great...

simple resizing to 3840 using sinc algorithm:

https://www.youtube.com/watch?v=H4hS_uBehH8

After shooting this video at a little over 3400*1700, I simply used 2x superscaling (sharp, sinc) in resolve to easily and quickly bring it up to 3840 with beautiful results!  Take a look at the beautiful sharp mesh around the subject tree in the opening scene...

https://www.youtube.com/watch?v=lO1I4UJr2kE

There was a sudden stop during the second shot, and although I was only 6 feet from the camera, I couldn't hear those few short quiet beeps in a forest full of singing birds, and I kept talking without knowing the recording had stopped.

Can someone PLEASE add an option for a continuous tone when the camera suffers a stop?

I asked years ago, and was ignored.  I know it's fashionable to ignore me, but is there really no one out there who also needs a better alarm when the recording crashes?

really?

This was shot in the forest, and I know from experience that it would have stopped suddenly if i was using the 3.X "x5" modes.

I'm wondering if the uhd 1:1 mode is more reliable because of the better data balancing between the two cards?

I find it to be almost equal data written to the SD and CF cards.

I'm really liking the look I'm getting in this resolution, including the DOF!

https://youtu.be/aPTMbGr1g-I

I noticed during a recent video shoot that 3840*1536 in UHD 1:1 mode seemed to be quite stable for recording with card spanning, overclock and bilal's hacks enabled, at least for a short clip.

Here is a video using only that mode.  It worked perfectly!

One thing to keep in mind:  This mode is quite good at balancing the data between the 2 cards, so you either need a 256GB SD card, or be prepared to run out of recording time pretty quickly...

No upscaling!  MLVFS straight to resolve.  Edited, graded and output to NVenc HEVC 10 bit 4:4:4 in record time!

I never thought I'd see native 3840 raw video from the 5d3...

https://youtu.be/t16bqfoqhbY

OBVIOUSLY watch in 4k!

Since Danne's suggestion that I should use separate card readers for MLVFS using card spanning, I've been doing every video in at least 3000 pixels across, and having zero problems!

Lately I have gotten lazy (and impatient) so I have switched from "superscaling" in resolve (GPU intensive and slower, but with beautiful results) to simple upscaling using the SINC algorithm.  The issue with sinc is how it strongly emphasizes any aliasing in the source footage (very badly, actually to the point of unusability).  But it seems that this only happens when the source footage is below a certain resolution, for I have never seen it at all in these latest videos.

This video was shot at 3072*1536 (I really like 2:1 aspect ratio) and upscaled to 4k using simple SINC.  To me it looks like it was shot at 4k. 

speaking of 4k, make sure to view it at 4k to avoid youtube quality degradation...

https://youtu.be/vNrEmilbvdw

I can't praise MLVFS highly enough.  To pull the cards from the camera and have a completed edited graded video uploading minutes later is pretty amazing!

PS: the banner image for the video is a frame from the video, cropped 200%

the peach blossoms look astonishing, just like real life!

5d3 3.5k raw - mlvfs - davinci resolve - cuda hevc 10 bit 444

1 hour from start to finish.

https://youtu.be/YgpW2Z5Z-Tk

Since danne's suggestion to use 2 card readers, I've done several videos using card spanning with mlvfs for instant grading/editing in resolve, with absolutely no issues!

the latest example:
https://youtu.be/o9w99l9zFgc

workflow:  stick cards into readers, create 32MB ntfs ramdisk, run "mlvlinks" script, run mlvfs script, jump into resolve!

see this thread for details:
https://www.magiclantern.fm/forum/index.php?topic=25473.msg243387;topicseen#msg243387

I shot another video in the forest, this time at the highest resolution i could get: 3664x1534

i was hoping that updating dokan solved the problems with mlvfs, but it failed even worse now, with crashes of mlvfs and/or dokan resulting in the virtual drive disappearing (as well as the command prompt window for mlvfs).

i tried updating to the very latest version of dokan, and renaming the resulting dll so it would work with mlvfs.  it worked, but didn't improve the situation.  in fact, it made it worse, by adding corrupted frames back into the mix...

in the end i got a perfect result by reverting to the last 1.x version of dokan, and then copying all the *.m000 files to the CF card

so it seems to be related to reading both cards simultaneously.  whether it is a driver issue, dokan, and/or mlvfs is unknown.  it could even be related to the actual card reader (kingston).  who knows...

hopefully others will explore this, particularly now that sd overclocking is so reliable.

again, in the end it worked perfectly by copying the footage files from the sd card onto the cf card (in the same directory)

it was simply amazing to shoot continuously at this resolution with no problems!

check out the exquisite detail of the footage.

watch in 4k!

https://youtu.be/RN4WaEOJonA

for my latest 5d3 video i decided to go for the highest possible continuous recording resolution (over 3500 pixels across) for maximum detail.

i was surprised and dismayed to find that at this resolution mlvfs makes errors every time i render in resolve.  i've done over a hundred long form videos using mlvfs and resolve at 1920 superscaled to 4k and it has never dropped a frame ever until now.

there were problems in the past, but they disappeared as soon as i switched from windows 7 to 10.

it most likely is because of timing difficulties with card spanning.  the CF card is much faster than the SD, even though it is supposed to be fast.  (sandisk 170)

maybe it would be improved with using 2 card readers on separate buses.

but it should be more robust than this.  it seems to be always on the edge of failure...

the glitch happens in the opening shot when i begin panning right:

https://www.youtube.com/watch?v=jJLPWZG8-ww

on a happier note, check out how absolutely beautiful the video looks at 4k!

it's a shame it suffers from the frame corruption because otherwise this is dreamlike in the convenience of pulling the cards from the camera, throwing them into the card reader and going straight to grading and editing and rendering in resolve...

i came in from the forest, and had a completed rendered copy ready to upload in under an hour, including recording the dialogue.

EDIT: upgrading to the last 1.x version of dokan seems to have solved the issue... v 1.5.1.1000

i re-rendered the video twice and both times it was flawless.

i'll do more videos at this resolution and hopefully it keeps working perfectly.

EDIT:
An additional note regarding the 5d3:  since it is now capable of such high speed data writing, 1920 probably wouldn't require data compression, freeing up CPU resources for resampling.

---------

I intended to shoot this video at 3616x1536, but there was a problem, even with card spanning and sd 240mhz overclock...

because of the bare branches on all the trees in the forest, the scene is just too complicated for that resolution, even when underexposed!

i had already shot the "beauty shot" for the middle of the video of the baby chestnut in the morning sunlight at that resolution, but when i got immediate sudden stops repeatedly in the forest i decided to shoot this talk at 1920 for reliability, using 3x3 mode since i was shooting at f/2.8 so there would be no problem with aliasing and moire on all the branches of the trees.

i wanted maximum detail on the closing shot of the cloudy sunset through the trees, so i shot that using 1x1 mode.

i was really surprised when i was editing the video to find that there was so much aliasing on the 1x1 shot!

to see the 1x1 footage, skip to 8:50 and pause it before the messages pop up for other videos.  the problem is very clear in the upper left corner of the screen.

i used 4x supersampling in resolve with gaussian sampling to minimize the problem, but it is still bad.

seeing this made me wonder if it is possible to do better resampling (even simple bilinear) of the full sensor down to 1920 instead of the horrible binning.  i know that it is beyond the capabilities of the cpu in most of these cameras, but has anyone done any experiments with this at all?  maybe certain dual cpu cameras could do it?  it makes me wonder if this could be done well in the very recent crop sensor models that have a newer (faster?) cpu.

it's just a thought...

on another note, if anyone is interested i could share a full sensor resolution still of the extremely challenging forest shot to use for a "stress test" of card overclocking.  if you print it out at full resolution, it could be useful as a "standard" torture test.  there was no wind at all in the forest and the camera was on a locked tripod, so a still should be just as difficult, especially with perfect exposure.

view the video at 4k, of course.
https://youtu.be/A8elVCzKfr4

with the renewed debate about 3k, i decided to shoot my latest forest video in 3k, utilizing resolve's superscaling algorithm to bring it up to 4k using 4X upscaling:

must view at 4k to see it correctly!  check out the incredible smoothness of the razor sharp spider webs that are clipping in the sunlight...

https://youtu.be/UA_o5qI5iqQ

ps: i'll send some frames to ml700d

IS was a lifesaver for a handheld focus pull at maximum closeup on the tiny redwood bud!

watch in 4k to bypass youtube quality degradation...

https://www.youtube.com/watch?v=N_2dI7dSzEc

The winter that wasn't...

Long point, Ontario - February 20, 2023
View in 4K for original quality.

https://www.youtube.com/watch?v=GtkoZhF2P5c

sometimes you want as much dynamic range as possible, even in very dim light, and with the 5d3 that means iso 200...

check out the beautiful detail on the LEDs in a nearly dark room, with the shadows severely lifted in post.

shot in 3616*1536 continuous (thanks bilal!), and upscaled to 3840 using sinc algorithm in resolve.

watch in 4k to avoid the horrible quality degradation of youtube imposed on videos with <3000 views.

best wishes to everyone!

https://youtu.be/M6U32od88EY

this is useless to us, but it is very interesting!
https://pub.towardsai.net/stable-diffusion-based-image-compresssion-6f1f0a399202?gi=2dde7fc8cea1

seriously dim lighting except the vocal booth, which unfortunately was a little too bright!

https://www.youtube.com/watch?v=L9fS2DtvCj8

I'm now making the occasional Blender3D tutorial video.  This one (the second in the series) covers workarounds for the "eccentric" rigid body simulation to make a chain that doesn't break:

https://www.youtube.com/watch?v=Jdy68qLdOR8

A short but sweet exploration of the forest in early morning.  The extreme dynamic range isn't even apparent in the resulting video because it was so easily controlled!
Every shot has both clipped highlights and lifted shadows, sometimes by many stops (such as the closeup of the chestnut tree bark).
Watch in 4k!
https://www.youtube.com/watch?v=I2LpW7Ofn90

timbytheriver's dolby vision/rec709 trick with resolve made dealing with the clipped flames in the last shot of this video not only better looking, but far easier to deal with.

in the original grade, i had to resort to trickery such as adding another node to isolate the clipped highlights to desaturate them and add back a false colour.  not only was it no longer required but the natural colour actually looks good even though it is clipped...

https://www.youtube.com/watch?v=KiJKKruptFM

pretty solid low light high dynamic range...
i re-rendered the closeups of the animated mouth, taking advantage of some of the new features in blender.  the differences are obvious between the closeups and the prior version in the camera-captured scenes:
https://www.youtube.com/watch?v=RIecXsKwFXc

Has everyone seen this paper from 2016?
I'm really curious about thoughts regarding this, particularly the performance of the ML method, from the upper echelons...
Dual iso has so much promise.  It requires a new approach to achieve any of its potential.

https://repositori.upf.edu/bitstream/handle/10230/28210/Bertalmio_ElectronicImaging2016_high.pdf?sequence=1&isAllowed=y

I thought this was already as good as it could get, but again, dolby vision gives just a little bit extra!

it was a little more difficult to handle the flames, as there were some vertical lines in them, but it was easy to fix in resolve by using a node that was keyed to them and applying blur on the red channel only.

watch in 4k!

https://youtu.be/b8BAGnapTqQ

after having graded a bunch of videos with dolby vision, i've created a power grade for resolve that incorporates everything i've been doing manually, which actually isn't that much...  mainly sat vs sat and lum vs sat curves and a few other adjustments on the wheels.

i applied that power grade to this video, and did nothing else:

https://youtu.be/KYICJdELcPs

I've been re-grading more videos with Dolby Vision, and the results are fantastic.
This video was shot in low light, and the richness of the colour palette is phenomenal, with no banding to be found, and grading with Dolby is so easy!
Basically it only requires careful adjustment (rolloff) of the lum vs sat and sat vs sat curves on the right side of the curve. I'm guessing it is an issue somewhere with how resolve reads MLV DNG files?
It's not too difficult to adjust.

https://www.youtube.com/watch?v=7fskKhCkpYg

effortless grading with tim's dolby vision trick!
all i did was roll off the saturation on the highs and pull back the highlights to exactly 1000 nits on the HDR scopes...
the closeup at the end is breathtaking.
watch in 4k!
https://youtu.be/g8Qb5PsFnDc

timbytheriver came up with a method for hybrid HDR-REC709 processing with Dolby Vision, and it made a dramatic improvement to this example video I shot a few years ago on my 5d3 (3072*1308 14 bit)

I previously had serious difficulties with the more extreme dynamic range shots, particularly the closeup of the oscilloscope.  It formerly required trickery such as spot exposure reduction on the LED to get the right saturation without losing all the contrast in the rest of the shot.  Now it was straightforward and looks great with no trickery!

This workflow also revealed details that were previously absent, such as the sky through the (filthy) windows in dimly lit interior scenes.  Incredible colour!

On another note, I was so pleased with the new grade that I re-rendered the blender animations, and in the process I found that the "eevee" realtime raytracer is now very close to production ready, so I used it, and it was a real game changer.  What a treat to see lighting adjustments on the fly...

https://www.youtube.com/watch?v=gHZ2MdPFtLw

timbytheriver suggested I should try using mlvapp to do basic exposure and colour temperature adjustments and then use the arri preset to export for editing and final grading in resolve in arri log c (using the CST plugin).

I gave it a shot for this music video and I'm hooked!  The fantastic mlvapp colours make it into resolve!

Shot in 14 bit raw on my trusty old 5d3 @ 3616*1536 using card spanning and overclock. (except the cellphone footage of the guitars!)

https://www.youtube.com/watch?v=r3-6cUuVTL0

Best wishes to everyone.

https://www.youtube.com/watch?v=lOgv7_CuCv8

The opening scene of this video is by far the darkest scene I have yet pulled off, far darker than the candlelit scene (which had nothing but 6 candles).
ISO 1250, shot at 3616*1536, 14 bit lossless with card spanning and overclock.  NO noise reduction, exposed for the fiber optic light curio.

The rest of the video is dual ISO with much more light.

Watch in 4k in a very dark room!

Read the video description for more info.

https://www.youtube.com/watch?v=-aE-HSdVtYk

This seems to be the limit of what can be done with the 5d3.
I'd like to try the sensor calibration that has been stalled for years, and I'm willing to risk frying my sensor!
(remastered)
https://www.youtube.com/watch?v=05_vPCkJLmY

I've gotta say, ACES 0.1.1 in resolve 17 is pretty amazing for magic lantern raw footage!

watch in at least 1440p for no quality loss.

5d3 14 bit

https://www.youtube.com/watch?v=BMIGFDE8sVk

shot by myself and timbytheriver on our 5d3 cameras, one on each side of the atlantic ocean

https://www.youtube.com/watch?v=E5q0YmwXJ3Y

shot in 14 bit 23.976 fps 3840x1536 using card spanning and overclock.  the candles are the only light source.

https://www.youtube.com/watch?v=xIPtOkM5HaI

The new debayering algorithm in resolve 17 works wonders with 5D3 3x3 footage!
It's so great that I no longer use other modes.
Real time correct preview is another treat!
Everyone with a 5D3 needs to try this.  I hope someday mlvapp will be able to debayer this well!
Seriously, this is a whole new lease on life for the 5D3.  I wonder if the 5D4 has a similar 3x3 mode?  Imagine what it could do with that improved sensor...
https://www.youtube.com/watch?v=0tEkCGJ3dZM

I wasn't 100% convinced that DWG was beneficial for SDR video, so I shot this extremely challenging scene and it really makes a difference!
When the clouds roll in, it stays beautiful.  This is static processing: one setting for the entire scene!
https://www.youtube.com/watch?v=Q0ElPjBh9bQ

this is my swan song from magic lantern.

if anyone wants to keep in touch, please subscribe to my channel.

best wishes to everyone,
farewell

https://www.youtube.com/watch?v=G7egUYMWeHc

With the talk about youtube performance, I thought I would give it a serious test using TNT at 1440p:

I shot this at 2560*1440 (14 bit using TNT) and graded it in resolve to put all it's got into the darks.

Watch it in a dark room on a good screen at 1440p!  Look closely at all the textures and detail of my black shirt under the coat.

As always, no noise reduction of any kind.  TNT at its best!

Encoded using NVENC h.264 @ 30MBPS

https://www.youtube.com/watch?v=T3oikPHyUNY

in honour of the 40th anniversary of getting the synth that changed my life, a rich dark punchy celebration:

https://www.youtube.com/watch?v=Tx5x3VFf6vU

if it wasn't for the subtle vertical stripes noise in the bright midtones (such as piano keys) I would call this perfect looking...

https://youtu.be/RgQRxN2CQic

windows resolve users can now work with card spanning straight from the cards!

i combined the "lossless" version of mlvfs.exe with a newer version of dokan, and packaged it with my little batch file for "linking" 2 cards into one virtual drive, and added a readme file to provide step-by-step instructions.

once you are set up, you can take your 2 cards out of your 5d3, put them into your card reader, and resolve is ready to rock!

update (feb 2023) : Dokan version 1.0.3.1000 is recommended

zipped for your convenience:
https://1drv.ms/u/s!ApLIbwc4O9qTcXltp8brBpxcJcg?e=agxYB7

5.7k raw 5d3
https://www.youtube.com/watch?v=mbvqcbmzrO4

when i was grading/editing this video in resolve, i ran into a nasty bug...

my 5d3 has a stuck pixel, almost dead centre of the sensor (of course!) and i always take care of it with the "dead pixel fixer" plugin in resolve.  the plugin stopped working when i loaded this footage, and i went looking online for any hints as to what was wrong.  luckily, the search was good, and i found it happens when there is resizing at the "clip" level, which is what i was doing.

so i disabled the resizing, and changed the project settings to the exact desired output resolution (in this case 3840x1584), with "mismatched resolution files" resizing set to "stretch to all corners":

i set the timeline to match.

when rendering, set the resolution to 3840x2160 and ignore the warning.  it will then add the black bars to the render.

the plugin worked again, and as a bonus, playback was much smoother - realtime, in fact!

hope it helps someone...

https://youtu.be/zQAj0MBJwaA

this was originally posted in the mlvapp thread, but i just noticed a new user who didn't know about this, so it seems like a good idea for it to have a separate thread for easy discovery:

i wrote a very simple batch file for other windows users to use:  (input it as a text file and then save as filename.BAT)
right click on the icon and choose "run as administrator"

Code: [Select]

@echo on

SET SrcRoot1=L:\DCIM\100EOS5D
SET SrcRoot2=S:\DCIM\100EOS5D
SET TargetRoot=A:\MLVSDCF

FOR %%A IN ("%SrcRoot1%\*") DO (MKLINK "%TargetRoot%\%%~NXA" "%%~A")

FOR %%A IN ("%SrcRoot2%\*") DO (MKLINK "%TargetRoot%\%%~NXA" "%%~A")

PAUSE
EXIT


it's obviously set for my paths, so adjust to your specific settings.  make sure to create the destination directory before running the batch!

i'm using a ramdisk as drive A, which is a convenient place to stick random junk like a pile of shortcuts for a single use

NOTE: the destination directory must be formatted with NTFS

shot on both sides of the atlantic, using TNT of course

it only made sense to call our music project TNT also!

magic lantern brought us together to hack our cameras and make music too!

the song was inspired by the sudden catastrophic loss of all the butterflies in my forest after the farmer next door poisoned his fallow field in late summer.  i'm playing my keyboard on the boundary between my forest and his field.

https://www.youtube.com/watch?v=-T65_RjCfJI

i sometimes use 18fps when i want a different cadence, and i find it suits this kind of video.

maxed out resolution @ 2:1 AR, recording strictly to CF, 14 bit:

there's definitely extra efficiency in danne's 3.x k modes.

https://youtu.be/RAYQ0llEFBM