Messages

@ayshih: Thanks for your explanation, so I should use Samba server. I will try, but I never did this yet, so it will be a long journey for me. Cygwin is not for this job.

I could install Ubuntu (14.04.1) inside the VirtualBox (test build 4.3.15) but the MLVFS isn't working. I think it's because I'm trying inside a virtual machine, so there is too much virtualisation "layer" for the system. Now I will try the Cygwin (never used it), maybe it will work.

Update: I could get it working with the new stable 4.3.16 VB. It's a very long journey to do, but it works. As I have time I write a step by step "how-to".

I will experiment with the Linux (Ubuntu 14.04.1) in a Virtual Box on Windows 7 in the evening. It should work in theory, I will report back:
1. Run the Linux in Virtual Box on Windows 7.
2. Run the MLVFS in the virtualized Linux system.
3. Open the DNG sequence in Windows 7.

Update: I can't get it working, the latest stable (4.3.15) Virtual Box drops error for me, but I'm on that to solve somehow.
Update2: I could get it working with the new stable 4.3.16 VB. It's a very long journey to do, but it works. As I have time I write a step by step "how-to".

I just updated the OP with my settings from remembrance.

[Don't initialize!]

At first save your ML generated sidecar XMPs as backup somewhere to be safe. If you do something wrong, you can start the process again.

It was very long time ago, when I shot that timelapse but I will try to remeber how I did it.
1. Make a folder, copy the CR2s and the ML sidecar XMPs into it.
2. Open the Lightroom, import the folder.
3. Open the LRTimelapse, navigate to the folder.
4. Switch to the Holy Grail Workflow tab. Don't initialize! This is very important, because you have information in the Exposure 2012 column, they are from the ML sidecar XMPs.) -> Click on Keyframes Wizard (Click yes if it asks about ISO + exposure jumps) -> Click on Holy Grail Wizard -> Adjust the offset + rotate bar if you want. -> Click save XMPs.
5. Follow the Holy Grail Workflow.

But I think you should try dmilligan's ramping and deflickering Bridge script too.

@dmilligan: awsome thought.

I know it's too early to write about it, but I think it would be also nice to add and option to apply DNG Camera Profiles (.dcp files) too. I mean the index file should contain the information, that during the virtual file system upbuilding which .dcp file should be applied to the DNGs. There is a utility, the dcpTool which could be a good starting point if a xml file is needed for this.

I have a motorized dolly to get motion when I'm timelapsing and I just thought about something. Would it be possible to trigger the flash shoe after that ML saves the DNG into MLV?

It would be usefull to trigger the MX2 motion controller to move the dolly, please read the comments here and you will understand what I'm speaking about.

You have the module in this previous dmilligan post: http://www.magiclantern.fm/forum/index.php?topic=12733.msg122887#msg122887

But be aware than you need also a complete ML build for your camera from the fullres-silent-pics branch. You cannot put the silent.mo in a regular Nightly Build, it will not work.

Ask for a fullres-silent-pics build for your camera here: 
http://www.magiclantern.fm/forum/index.php?topic=12608.0

Thanks for the link to the dmilligan post and for the warning.

I have this 550D nightly build created by mk11174
(Mercurial changeset: 25735947ce04+ (fullres-silent-pics), built on 2014.07.18.

Is this ok for testing purpose?

Yes I know about that topic, I asked there too for the proper build.

May I have the latest silent.mo module to test on the 550D, I only have to previous one without the file format options.

May I have a nightly build for 550D with the latest full-res silent pic module to save the frames to MLV? I would like to test it.

Would it be possible to send the frames from the SRM job memory buffers to the SD or CF (camera specific) card and to the USB port (maybe with Wi-Fi read out) paralelly?

Edit: ... or split the frame and send the top half to the SD or CF card and the bottom part to the USB port. Later in post process merge the 2 parts to 1.

Awesome job a1ex! Could this function work together with ettr + deflicker + advanced intervalometer module too?

Yepp... sadly more HDD is the solution or to find someone, who can write a 16 bit frameserving plugin for AE/PP/AME. If it is solved, than PIPE -> PIPE -> PIPE

Maybe this things are helpful:
Instructions on how to setup and build the Debugmode Frameserver: https://code.google.com/p/frame-server/wiki/BuildInstructions
Premiere Pro CS6 SDK: http://www.adobe.com/devnet/premiere/sdk/cs6.html
VapourSynth: http://www.vapoursynth.com/about/
Raw format reader for VapourSynth: https://github.com/chikuzen/vsrawsource (see format_list.txt)
VapourSynth -> ffmpeg: http://forum.doom9.org/showthread.php?t=167627

Could you turn my head into a right direction, that how to write this kind of Bridge script (never did this)? Do you think, that there is a simple solution to watch the RAM Disk contuniously. I'm thinking about something like this: If the amount of the TIFFs is more then 200, delete the first 150 and after a little time check this again, again and again and delete if the limit is reached. So at the end I could have a continous video, not pieces. This way in theory the ffmpeg could run continously.

What is, when only so many 16 bit TIFFs are exported by Adobe Bridge at once what is enough to feed the ffmpeg to get the information be converted. With a delayed exporting and deleting cycle it could work or not? There is the RAM Disk, it's very fast, so this would be the buffer. Could this be solved somehow?

It's too much storing space to convert the CR2s + XMPs to 4K 16 bit TIFFs. Why isn't there any solution for this in 16bit, I mean frameserving? I don't understand. This could automate the whole process to get a final 4K 10bit 444 ProRes (or any other formats) file at the end on Windows, the RAM read/write speed is also faster than HDD or SSD.

On a mac here, sorry. All I do is select ProRes 444 in AME

I thought...

Imagesource?

It makes only 8bit processing.

I'm just reading this and thinking (#22):
http://forum.doom9.org/showthread.php?t=162598&page=2

What would be if the CR2 + its XMP datas (adjusments) could be converted into a virtual 16 bit TIF container (RGB48), but at once only 1 frame. This TIF file would be read out by ffmpeg and converted with prores_ks encoder, so we are at a 10bit frame (YUV 444) and this would be saved in a MOV container. If the 10 bit frame is already in the container, then the cycle should be started again with the next CR2 + XMP data pair. With delayed programing it could work not or is this bullshit? Am I thinking wrong?

Dcraw can read out the CR2 + its XMP data and convert into 16 bit TIF?

I forgot to mention, that with 8bit 4K JPGs I produce 4K 8bit 444 MP4 video this way:

1. Import the JPG sequence in AE, sRGB + 8bpc, save project file, close AE
2. Put the project file in queue in AME, start frame serving with the Advanced FrameServer, RGB24 preset (video only), this is the "frameserver.avi"
3. In the AviSynth file (frameserver.avs) I have this:

Code Select Expand

AVISource("E:\timelapse_rendering\frameserver.avi", audio=false).AssumeFPS(24,1)
ConvertToYV24(matrix="Rec709")
4. Start the 4k_8bit_444.bat file:

Code Select Expand

ffmpeg -r 24 -i "frameserver.avs" -aspect 16:9 -s 3840x2160 -c:v libx264 -preset slow -crf 13 -profile:v high444 -pix_fmt yuv444p -x264opts keyint=24:min-keyint=1 -g 24 -filter:v unsharp=5:5:0.5:5:5:0.5 -x264opts colorprim=bt709:transfer=bt709:colormatrix=bt709 -r 24 "4k_8bit_444.mp4"
pause
5. As the rendering is finished stop frame serving, close AME and watch the produced video

Is there a 10bit frameserving solution somehow, because if it exists, then with AviSynth high bit-depth and ffmpeg x264-10bit codec we could produce 4K 10bit 444 MP4 or with ffmpeg prores_ks codec 4K 10bit 444 MOV.

I'm on Windows and I have got a display, which is 8 bit native + 2 bit dithered.

I tried the Advanced FrameServer, but it's only serves RGB24, RGB32, UYVY and YUY2, so 8bpc:
http://sourceforge.net/projects/advancedfs/
http://wiki.multimedia.cx/index.php?title=RGB
http://wiki.multimedia.cx/index.php?title=YCbCr_4:2:2

I would like to have 10bit 444 at the end.

Somebody knows a way to convert CR2s (14bit) + their XMP datas (adjusments) directly to 4K ProRes (10bit, 444)... of course without colorshifts?

If you don't care about the images being exposure-matched in overlapped area (I assume you don't, otherwise you would have used M mode and --same-levels), just enable Post Deflicker, and meter each image as you wish (Av or ETTR).

Of course, in this case, the recommended way to use ETTR is on the SET key (and meter before each shot). The Always On mode should be used when the exposure is unlikely to change between shots (because it meters from the previous shot).

I tried this as a1ex suggested and it worked very well.

https://www.dropbox.com/s/mx8luiqxjb8tsh2/panorama_03_small.jpg (deleted)

Next time I will test the same, but combined with Dual ISO, let's see what happens.

Thanks for the "width" tag.

I also thought about the AETTR, but in this situation there is too much difference between the frames. If I start to shoot with AETTR always on, then in the 1st direction (from the 6 horizontal + 1 zenit) it will be fine after the sample shots (2/3 shots till get the right settings). The problem is there, that the panorama head turns automatically by 60 degrees after 1 shot, so the AETTR sample shooting sequence will start again and after the 1st sample shot the head turns again... and of course the next frame also different from the previous so everything will be bad. So it's a shoot-move-shoot head.

The shooting round would be fine till the 2nd direction, it looks like this:
0. (0 degree): AETTR always on, making the sample shots to get the right settings.
1. (0 degree): Start the panorama shooting sequence, result a perfect AETTR shot.
2. (60 degree): Head moved to the next frame, but totally other light conditions, so the AETTR starts again to get the right settings, but at the 1 shot head turns...
3. (120 degree): AETTR believes the it is at the 60 degree direction, but not. so again...

I don't know what to do, but I think there is solution somehow, that's why I started to experiment with middle shutter value.

Update: The best would be to fix an avarage AETTR setting somehow for all, and fix the Dual ISO too and shoot everything (all the 7) with same settings. Maybe this way:
1. AV mode to find the middle manually by turn around.
2. Find that direction where the expo comp slider shows 0.
3. Switch to M mode: AETTR always on, make the sample shooting sequence and fix the perfect one somehow and setup Dual ISO 100/1600. Can this be done somehow?

Something like this or I don't know. Maybe you understand what I think.

It bothered me yesterday, that I can't shoot a nice Dual ISO sphere panorama, so I tried it again at nearly the same light conditions and the result is quite good. (I think I didn't meter the light correctly the day before yesterday and I set up my camera wrongly, thats why there was a big dark part on that panorama.)

What I did now:
1. Whitebalance: daylight; picturestyle: standard, f8, ISO 100, in ML: Dual ISO 100/800
2. AV mode: metered (evaulative metering) the light in the room (turned around with the panorama head) in every direction, so this way I found the max. and a min. shutter speeds. I didn't touch the exposure compensation slider. (The day before yesterday it was -1EV.)
3. Switched to M mode: adjusted the shutter speed exactly between the max. and min. values (so this is the middle). There was 11 steps, so I originated from the metered min. and turned the wheel by 5 steps.
4. Shot the panorama.
5. Processed the cr2s with the lates cr2hdr-20bit without any command parameters.
6. I imported the dngs into Lightroom and made this adjusments: whitebalance: 5000; highlight: -100; shadows: +100, whites: -10; blacks: -10 and removed the cromatic aberration. I didn't touch the exposure. Exported to 16bit TIF, 300dpi.
7. Aligned the TIFs in PTGui and created the panorama as a 8 bit TIF.
8. I opened the panorama in Photoshop and adjusted the shadows by: 5% and pushed up the brightness to 21 and saturatoin to 10. After that I saved as jpg, quality 12.

Here is a small resized result.
(deleted)

Sorry for this big image here, how can I thumbnail it, I saw here at many post? (see #2064)

Thanks for test a1ex, now it's clear.

Yes, I will send the download link for you in pm as soon as it will be uploaded.