fastcinemadng

[megapolis]

It's a pity that it doesn't work for me even after reinstalling. I tried logging the installation of your program in order to then delete its remnants manually, including registry entries, but even that didn't help. After rebooting and reinstalling, the demo mode still remains, but the most important thing is that my presets remain in the export settings via ffmpeg, which means that the program files still remain in the system and do not allow the program to perform a clean installation and not work in demo mode. This is my guess. Perhaps other users will also encounter this. I really hope you fix it This will be the best thing that has happened to me since I met magic lantern.

Thanks for your info. Application settings are not stored in the registry now.
They are at C:\Users\<UserName>\AppData\Roaming\Fastvideo LLC.ini
Licence file is stored at C:\ProgramData\Fastvideo
1. Please check the software version. The latest version should be 1.0.16.0, build 1124.
2. Please check license file *.dat in the folder C:\ProgramData\Fastvideo. If you see such a file, please rename or remove it and run the software once more.

We don't remove licensing info after reinstalling, this is a feature, not a bug.
If the above info won't help, please send us your questions and screenshots either here or via PM.

[dream951]

Thanks for your info. Application settings are not stored in the registry now.
They are at C:\Users\<UserName>\AppData\Roaming\Fastvideo LLC.ini
Licence file is stored at C:\ProgramData\Fastvideo
1. Please check the software version. The latest version should be 1.0.16.0, build 1124.
2. Please check license file *.dat in the folder C:\ProgramData\Fastvideo. If you see such a file, please rename or remove it and run the software once more.

We don't remove licensing info after reinstalling, this is a feature, not a bug.
If the above info won't help, please send us your questions and screenshots either here or via PM.

Good evening!
I did everything according to your instructions and the program worked!
The export works without a watermark and restrictions, the speed on RTX2060 is impressive, a 30-second 5.7k video with a bitrate of 100 mbit was exported for a little more than a minute.
Thank you so much for your software!

[megapolis]

In the latest release (which could be downloaded from the site) you can find experimental version of defringe feature to suppress the consequences of chromatic aberrations. The feature still has some minor artifacts on the image borders, but it's working and pretty fast.

[DeafEyeJedi]

In the latest release (which could be downloaded from the site) you can find experimental version of defringe feature to suppress the consequences of chromatic aberrations. The feature still has some minor artifacts on the image borders, but it's working and pretty fast.

This is indeed promising. Thanks @megapolis for your contributions on this one.

Especially the fact that we can take advantage of our GPU's to transcode.

I'll definitely check it out for Windows OS!

BTW, when do you think we can test out a beta for Mac OS?

[megapolis]

Unfortunately MacOS support is not possible.
Our SDK for GPU-based image processing is available for Windows, Linux and ARM (Jetson).

[megapolis]

As we've previously mentioned, the performance of Fast CinemaDNG software in most cases is limited by input/output bandwidth, not by GPU model. From that point of view, the SSD speed for read/write could be the bottleneck for RAW processing on GPU.

We've recently done some benchmarks for DNG to JPEG conversion on GPU. We've utilized the same 8-bit 12-MPix DNG image and very simple pipeline: raw image reading from SSD and loading to GPU, WB, demosaicing with MG algorithm, DCP profile, gamma, conversion to JPEG 4:4:4 and quality 100%, storage to SSD. On the NVIDIA GeForce RTX 2080ti we've managed to process 3000 images per minute, which we consider to be pretty fast.

That performance was limited by our SSD with 600 MB/s read/write capabilities. This is 50 fps to process 12-MPix input images, and this is exactly the result we've got (3000 images per minute). Actually, processing on GPU was around 3 times faster, so the bottleneck was with SSD performance, not with GPU. GPU load was around 30% for one-thread application, and that's why we see great potential for further acceleration of RAW-to-JPEG transform on GPU due to faster SSD and higher GPU load.

[megapolis]

This is the latest release of the Fast CinemaDNG Processor software with the following updates:
1. Highlight recovery option on the GPU
2. Bug fix for JPEG header (EXIF section)
3. CUDA-11.7 support (so, new driver and PC restart are worth doing)
https://www.fastcinemadng.com/download/FastCDNG_x64_Setup.exe

[dream951]

Hi all!
I'm using the latest build. There was a full-fledged export to tiff for further color processing, for example, in 3DLutCreator, very convenient and fast! So far, I have not seen a faster post-processing and I don't think I will. Thanks, great job!

[megapolis]

The latest release of the Fast CinemaDNG Processor for Windows with the following updates:
1. Bug fix for Lossless JPEG decoder
2. Bug fix for Defringe module
3. Bug fix for Resize and Preview modules
4. CUDA-12.3 support (new driver is a must, 9xx GPUs are not supported anymore)
5. Software performance on GeForce RTX 4090 is two times faster than on 2080ti
Download link

[vastunghia]

I just came into knowledge of the existence of ZLUDA, for the rest of us poor AMD users. Just gave it a try (this boils down to zluda.exe FastCinemaDNG.exe basically) and

• the app launches, which is black magic already to me -- without ZLUDA of course it complains that nvml.dll was not found and refuses to launch
• however, tried both to open the demo project and a random MLV file, and in all cases I get

Code Select Expand

Error occured:
Cropping image for viewport bitmap failed
Code: 14 (FAST_INTERNAL_ERROR)
Not that I really hoped everything would go butter smooth, so no surprise. Just wanted to share with you.

Did you ever consider porting your CUDA code to AMD ROCm? Or maybe there could be a way to debug a bit ZLUDA / FastCinemaDNG interaction?

[iaburn]

Is playing or editing DNG a problem for anyone these days in Davinci Resolve? It's very fast in all my machines, including "low" specs ones

[vastunghia]

Would be nice to avoid the time- and disk space-consuming task of MLV -> cDNG. And MLVFS not an option for real-time with heavy footage (say 3.5K 14-bit compressed).

Would be a game changer at least for me in picking / discarding shots.

[megapolis]

Thanks for your info about ZLUDA project.
As I understand, it could work just for non-optimized GPU software. For example, if you've automatically ported any C++ project to CUDA, that could possibly work.
In the Fast CinemaDNG application we are utilizing our image processing SDK which is highly optimized for NVIDIA GPUs and it takes into account most of NVIDIA requirements and limitations for high performance software. We don't think that this software could work on AMD due to the differences in hardware and software architectures.

Did you ever consider porting your CUDA code to AMD ROCm?

Yes, we considered such an opportunity, this is very complicated and it would take years. Unfortunately we don't have resources to accomplish that.

Would be nice to avoid the time- and disk space-consuming task of MLV -> cDNG. And MLVFS not an option for real-time with heavy footage (say 3.5K 14-bit compressed).

Can't you do that with the current software? If it's not, please send me your MLV for evaluation.

[vastunghia]

Can't you do that with the current software? If it's not, please send me your MLV for evaluation.

"current software", in the context of the task MLV -> cDNG, for me means MLVFS or MLVapp. Just for clarity.

And no, no way to get real-time playback / conversion with 3.5K (which is 3.7K actually) 14-bit lossless compressed MLV raw footage. At least with my specs (super-fast NVMe drive + i5-8500).

Still interested into checking one of my MLV files?

Btw, thanks for your feedback on FastCinemaDNG vs ZLUDA! I had some hope since apparently it works out of the box even for high-level and (apparently) heavily Nvidia-optimized software such as Blender. But I guess it would take plenty of time anyway.