What's bottle-necking my editing speed?

The motherboard the OP mentioned is the Asus Z170A. The specs for that motherboard
support 4 lane PCIe M.2.
https://www.asus.com …ds/Z170-A/specifications/

A Samsung 950 Pro would help, but the Raid configuration is a large part of the problem.

Regards,

110yd

110yd wrote in post #18043704
The motherboard the OP mentioned is the Asus Z170A. The specs for that motherboard
support 4 lane PCIe M.2.
https://www.asus.com …ds/Z170-A/specifications/

A Samsung 950 Pro would help, but the Raid configuration is a large part of the problem.

Regards,

110yd

I see that, being too thick to understand RAID, I went for the quick and dirty solution.
After a shoot I down load it to my intel 750, from aLexar superspeed USB 3.0 reader. Its the fastest I found and going into the 750 reads at 120MBs, so acceptable .. 5 mins!
Then I copy it from the Intel to the Sammy 850 500GB . Very fast and then copy to the other Sammy 850. So I have 2 backups ( on the interim back up SSDs. And one working copy on the Intel 750. Once I edited some I put them in an "Edited folder " and quicky copy to both Sammys.

So all my working is done on my fastest PCIe SSD on. mB with 40 lanes.
As others have said I guess PSE13 is about as badly written as LR! In that most of the times it only uses 1 Core! Out of 6! But his CPU is faster than mine as mine is a 3.8GHZ xeon. Inthink his is 4.2GHz.
I Saw a huge increase in speed, and can visually see a difference in speed if I edit on the Intel 750 compared to the sata based sammy 850.
Like John_T my files are 53-90MBs but once you open them in PSE13 they magically expand to circa 160MB.
And moving them around on the intel and opening them is a dream. i guess the whole shoot is in RAM, as I have 32GB.
Plus as soon as the plugins are used then it leaps to 4 Cores, and really one sees a benefit.

And of course as there is no RAW viewer for the 5DS R one has to have DPP4 open at the same time , somits nice that really flies.

OK my solution aint raid and is probably a bit Heath Robinsonish, but at least I am working on the fastest bits of kit I own - and as you say ... Not a lot one can do until Adobe deliver well written programs.
Using the NIK collection also means more stuff can be done on decent software.

I just wish Google would get the NIK guys to write a Photoshop killer!

Submariner wrote in post #18043844
Like John_T my files are 53-90MBs but once you open them in PSE13 they magically expand to circa 160MB.

It isn't really magic at all. Your CR2 RAW files aren't really an image file, in that they don't contain colour pixels. They simply contain a number that defines the level of brightness recorded by each sensel, with no colour information recorded at all. When you convert your RAW data to a colour image, the software knows about the colour properties of the red, green, and blue filters used in the Bayer Colour Filter Array infront of the sensor and uses that information to build the colour that is seen in the image.

Each of the sensels brightness values is stored as a 14 bit binary number. Since a Byte contains 8 bits the simplest way to save that number is to put it in two Bytes, and just leave two bits empty. Or you could add complexity to the reading and writing stages and squeeze 8 sensel values in to 14 Bytes. If you go for the simple option you have to have 2 Bytes for each sensel location. For a 50 megapixel camera you would then need to have 100 Megabytes to hold that data. Canon use a lossless compression system, similar to what is used in zip files. This is why the size of the RAW file on disk is smaller than 100MB.

In PS or PSE in the pixel editing window (rather than the ACR RAW editor) the size that is quoted in the small box at the bottom of the window is the amount of memory required to hold the current image in memory. This though is an RGB image comprised of pixels. Each pixel comprises of a value for Red, Green, and Blue. You have the option of using different amounts of memory for each colour channel in each pixel. Using more memory for each channel allows for greater accuracy, and smaller steps between each possible colour. These days images usually have a minimum of 8 bits per colour per pixel, and quite a lot of people use 16 bits when editing. From the information in your post it is obvious that you are using 8 bit colour. That is not surprising, as PSE has very limited 16 bit colour support.

So each pixel uses 8 bits, or 1 Byte for each colour. So you will need a toatal of three bytes to hold a pixel. I'm not sure of the exact number of pixels in the 5DS, but 160 MB would be 53 MP.

If you were then to save the image as an uncompressed TIFF file this would be the size of the file. JPEG files, although having the same amount of picels, and the same 8 bits per channel, and so being the same size when open, will be between 60% and 40% smaller when saved to disk using the optimum quality settings, depending on the amounts of fine detail and noise in the image.

Alan

Maybe these links might be useful in determining what is what and what is good and what is less so:

https://photographylif​e.com …phy-needs-skylake-edition

https://tinkertry.com …0-pro-nvme-on-superserver

https://www.pugetsyste​ms.com …lti-Core-Performance-649/

BigAl007 wrote in post #18043943
It isn't really magic at all. Your CR2 RAW files aren't really an image file, in that they don't contain colour pixels. They simply contain a number that defines the level of brightness recorded by each sensel, with no colour information recorded at all. When you convert your RAW data to a colour image, the software knows about the colour properties of the red, green, and blue filters used in the Bayer Colour Filter Array infront of the sensor and uses that information to build the colour that is seen in the image.

Each of the sensels brightness values is stored as a 14 bit binary number. Since a Byte contains 8 bits the simplest way to save that number is to put it in two Bytes, and just leave two bits empty. Or you could add complexity to the reading and writing stages and squeeze 8 sensel values in to 14 Bytes. If you go for the simple option you have to have 2 Bytes for each sensel location. For a 50 megapixel camera you would then need to have 100 Megabytes to hold that data. Canon use a lossless compression system, similar to what is used in zip files. This is why the size of the RAW file on disk is smaller than 100MB.

In PS or PSE in the pixel editing window (rather than the ACR RAW editor) the size that is quoted in the small box at the bottom of the window is the amount of memory required to hold the current image in memory. This though is an RGB image comprised of pixels. Each pixel comprises of a value for Red, Green, and Blue. You have the option of using different amounts of memory for each colour channel in each pixel. Using more memory for each channel allows for greater accuracy, and smaller steps between each possible colour. These days images usually have a minimum of 8 bits per colour per pixel, and quite a lot of people use 16 bits when editing. From the information in your post it is obvious that you are using 8 bit colour. That is not surprising, as PSE has very limited 16 bit colour support.

So each pixel uses 8 bits, or 1 Byte for each colour. So you will need a toatal of three bytes to hold a pixel. I'm not sure of the exact number of pixels in the 5DS, but 160 MB would be 53 MP.

If you were then to save the image as an uncompressed TIFF file this would be the size of the file. JPEG files, although having the same amount of picels, and the same 8 bits per channel, and so being the same size when open, will be between 60% and 40% smaller when saved to disk using the optimum quality settings, depending on the amounts of fine detail and noise in the image.

Alan

Hi Alan
Thanks so much - great to learn something. Logic as usual beautifully explained.