C1 LE Very Slow on Fuji S5 files
CaptureOne LE takes a very long time to process Fuji S5 Files, compared to Nikon D70s, Canon 20D and 30D. On my laptop (1.6 GHz Pentium D) it takes about 75 seconds for an S5 file, versus 30 seconds for a 20D file. Obviously it is faster with faster processors (more like 50 seconds on a dual 64-bit AMD) but still much slower than other file types.
I know the files are much bigger, due to the extra image stored in the file, but it seems like it should be faster. I doubt if C1 even uses the extended range stuff in the file, so it is really just a 6 megapixel image, 25% smaller than the 20D.
On another performance issue, I noticed when processing images, C1 puts a very large (~37 MB) temp file in the develops directory while processing, then saves the ~3 MB jpeg. The develops directory is usually on a network drive, that is much slower than local hard drives. For improved performance, that temp file should be stored on a local drive, or better yet, in RAM. An extra 50 MB RAM footprint would be insignificant for most machines, and could significantly improve performance!
Is anyone else using S5 files?
Thanks,
Paul Samuelson
I know the files are much bigger, due to the extra image stored in the file, but it seems like it should be faster. I doubt if C1 even uses the extended range stuff in the file, so it is really just a 6 megapixel image, 25% smaller than the 20D.
On another performance issue, I noticed when processing images, C1 puts a very large (~37 MB) temp file in the develops directory while processing, then saves the ~3 MB jpeg. The develops directory is usually on a network drive, that is much slower than local hard drives. For improved performance, that temp file should be stored on a local drive, or better yet, in RAM. An extra 50 MB RAM footprint would be insignificant for most machines, and could significantly improve performance!
Is anyone else using S5 files?
Thanks,
Paul Samuelson
0
-
Paul,
Don't have Doubts ... all the chip data from the Fuji Pro cameras are utilized in Capture One 😄
The reading of data for the process algorithms is done in a horizontal fashion.
Processing is longer for the Fuji files because the diagonal chip data takes longer to read.
If Speed is more important to you, then use the QuickProof algorithm which is built with speed as the priority. The Process algorithm is built with Image Quality as priority.
The temp file you mention is created at the Process To location. It is a 'place holder' of sorts.
It is always best to Process to a local machine hard drive for efficiency.
I alway advocate the best advice to Keep It Simple.
Locate Session folders on local hard drives, process, and then move them to server locations.
The best setting to have established at the Process tab is to the Set Process Folder… and check the Always process images to associated image folder
With this box checked, a folder name Develops is created within the Session folder along with the RAW images.
Creating a Process to folder across a network can have problems depending how it is set-up.
I think it is best to keep the Process to location consistent.
By checking the box to Always process images to associated image folder accomplishes this consistency well.
Once processed, the folder can be renamed, copied or moved to the Server Location quite easily.
Sincerely,
K C0 -
Sorry I doubted your use of extended range chip data from S5 files. I am impressed (and surprised) that you use it, and it probably explains the performance compared to simpler files.
Speed is important to me, but not at the expense of quality, so QuickProof is not an option.
I have been reading here about a 300% speed improvement in v4.0 when it is released. Is that across-the-board improvement, or only with certain file types? Will it include S5? I signed up to beta-test, and look forward to trying it out as soon as I can.
On the temp file issue: I have the Raw files stored on a server, and when I am done processing, I need to have the jpgs there as well. I can certainly change my workflow to save the files locally, then copy them back to a server. (A quick test yesterday showed about a 20-25% improvement in performance).
However, this just confirms what I thought - the performance issue is due to writing/reading the large temp files, not the writing the processed jpg or tiff. If the temp files where stored locally, it wouldn't matter where the output folder was. Ultimately, I need to get them on the server anyway, so I don't care if it takes slightly longer to write them to the server. It would likely be quicker to write them there directly, than write them locally and them move them. I just want to create them as quickly as possible.
If you stored the temp files in a user-configurable local directory (like Photoshop's Scratch Disk) - or better yet, in RAM - the processed files could be saved in any location without a performance issue.
Thank you,
Paul Samuelson0
Post is closed for comments.
Comments
2 comments