RAW to JPEG

17 replies [Last post]
Pete Allen
Offline
Joined: Oct 25 2006

Can anyone tell me the best way to convert RAW images to JPEG, so as to lose as little of the image as possible (resolution and visible area) without distorting, e.g. when resizing to 8x10, 5x7, 6x4 etc.
Thanks

I drink to steady my nerves. Last night I got so steady I couldn't move. Wedding video essex

Alan Craven
Offline
Joined: Jan 26 2001

I use the Adobe dng Convertor, which is a free download, to convert my RAW files. the .dng can be opened in Photoshop for conversion or further editing.

foxvideo
foxvideo's picture
Offline
Joined: Sep 9 1999

Raw Therapee is worth every penny.....(it's free!)

Excellent conversion package with many of the tools of the expensive paid packages.

Dave Farrants Fox Video Editing

Chris.
Offline
Joined: Nov 5 2000

What camera are you using Pete?

Pete Allen
Offline
Joined: Oct 25 2006

Using a Canon 40d at the moment, with lightroom to transfer images.

I drink to steady my nerves. Last night I got so steady I couldn't move. Wedding video essex

Chris.
Offline
Joined: Nov 5 2000

Pete, Lightroom is a good RAW converter, personally I like Canon's own Digital Photo Professional.

Either one should give you very good JPEGs from the RAW files.

If you're having quality issues with your JPEGs there's a few common mistakes that you need to check.

Don't save the JPEG with an unsuitable colour space such as AdobeRGB. Ideally save JPEGs in the sRGB colour space, this is much better for applications like email and displaying on the web. Many web browsers would display an Adobe RGB JPEG very badly, muted colours, loss of contrast. Though printers print with CMYK inks, the printer driver expects RGB input. Again use sRGB.

The aspect ratio of your camera is 3:2, this is a direct match for print sizes such as 6 x 4 and 12 x 8. It's not a match for other common sizes like 10 x 8. If you need to print at 10 x 8 then you should crop the image. If you didn't crop the image it would have to be squeezed (distorted) to fit, or a border left around the image (The same issue that widescreen and 4:3 TVs have)

The other thing that causes problems is a misunderstanding of PPI and DPI; people with good intention altering the setting and ruining their images. Check this page http://www.rideau-info.com/photos/mythdpi.html

What particular problems are you having?

Rob James
Offline
Joined: Jun 26 2001

Chris, that link was extremely useful and has demystified something that has been worrying me for a long time.

Many thanks!

Rob The picture is only there to keep the sound in sync

Chris.
Offline
Joined: Nov 5 2000

Rob I'm the first to admit that I once struggled with the concept. At one time getting mixed up with dots per square inch.

Now even I find it pretty easy to work out that a picture 3000 pixels wide will print 10 inches wide at 300 ppi! And so on

Printer manufacturers don't help when they claim 1440 dpi resolution for their machines, but when you just think of that as drops of ink per inch, and that it takes several ink drops of different colours to reproduce one pixel it starts to make sense.

Rob James
Offline
Joined: Jun 26 2001

Chris, I've been playing with digital images for at least twenty years and that is the first explanation that made sense so I'm deeply indebted to you. I had worked out the ppi relationship with print size but the others had me flummoxed. I'd still like to know why 96dpi images work well at 100% in PDF documents while 72dpi ones don't but, other than that I'm a lot less mystified than I was.

Rob The picture is only there to keep the sound in sync

Chris.
Offline
Joined: Nov 5 2000

It's possibly a DTP thing, many DTP packages read the dpi tag in an image file to help them decide how large the image should appear on a page.

The other possibility is that monitor dpi (more correctly ppi) is playing some part, though I think that's redundant these days. When everyone had either 15 inch or 17 inch monitors and ran them at either 800 x 600 or 1024 x 768 then monitor dpi made some sense.

Some of the new micro laptops have very small but high res screens, I think with 150ppi resolutions. An application would only know the screen res but not on what size monitor it was displayed on.

LOL I'm thinking out loud now. Short answer, I don't know why PDFs do that

Rob James
Offline
Joined: Jun 26 2001

Thanks for the thoughts Chris. In the "day job" I write software manuals (amongst other things). For the manuals I use FrameMaker and I long ago discovered empirically that 96dpi worked for screenshots etc. and 72dpi didn't. I'd love to know why but, "if it ain't broke, don't fix it" applies.

If the resultant pdfs are viewed at 100% the screenshots look good on various resolution screens. At 72dpi they look horrible.

Rob The picture is only there to keep the sound in sync

Chris.
Offline
Joined: Nov 5 2000

Out of interest Rob are the real world pixel sizes of the pictures identical and the only thing you've changed is the DPI properties?

What happens if you use an even higher DPI, like 150 or 300?

Rob James
Offline
Joined: Jun 26 2001

The point at which "DPI" comes into the equation is when the pictures are imported into FrameMaker. The value chosen does affect the size of the image in Framemaker and when I began using it I simply sized images by using the import dialogue and selecting a suitable "DPI". (The images are mostly screenshots) I was often disappointed with the resultant quality so I did some empirical experiments. I obtained MUCH better results by re-sizing in Photoshop and importing into FM at 96DPI. Given that these documents are intended to be read on screen rather than be printed I suspect the reason this works is down to Windows using the same value regardless of the display pixel resolution. I.e. If one of my pdfs is viewed at 100% then the images are a pixel for pixel match with what I see here.

Rob The picture is only there to keep the sound in sync

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

Windows seems to have two screen resolutions, 96 and 120dpi. The actual dpi figure for the display is never used in the software, but I have a Dell laptop (1920x1200, 15.4") that says it's 120dpi (it isn't) and a DVC special (1920x1200, 17") that says it's 96dpi (it isn't).

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

Rob James
Offline
Joined: Jun 26 2001

I think this is the nub of the matter. So-called 96 DPI is I suspect, the most commonly found varient. However, I don't believe this has anything to do with dots per inch at least not in an obvious way. I reasonably convinced that the reason my pdfs look best using "96dpi" is simply because you then get a 1 to 1 pixel mapping. (Providing you view the doc at 100%) I.e. I'm making screen captures on a system using "96DPI", whatever that means, and so long as I don't re-size the screenshot and use "96DPI" in Framemaker then when the pdf is viewed the screenshots are identical to the original. I've an idea that Macs use 72DPI or used to.

Rob The picture is only there to keep the sound in sync

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

Correct, 96 is the normal setting.

But, whatever the dpi setting is for a picture, if you choose to show it at "100%", you should see it pixel-mapped to the display, no scaling. But it wouldn't surprise me to find alternative interpretations of this.

You should be able to dive into the properties of any image file and change the dpi setting without changing anything else. For example, in a Windows BMP file, there are actually two numbers for dpi (one for horizontal, one for vertical) which define the number of dots per metre (not dpi), at offsets of 38 and 42 bytes into the header, plus two further numbers for the width and height in pixels, at offsets 18 and 22. It makes sense to do it this way, because when working with image files, you need to know the pixel dimensions first. And it gets interesting at this stage, because 72 dpi (most common setting in years gone by) doesn't work out at a sensible value in dots/metre, so you can set the dpm value to zero and it gets interpreted as 72dpi rather than a dpm value. I guess that other file types are similar, but I've never dived into them to find out. Complicated or what?

I had to come to terms with all this when writing my camera/display analysis software. Hours of fun ;)

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.

Rob James
Offline
Joined: Jun 26 2001

Aaargh! Fun or what?

FrameMaker uses the DPI setting in image files (generally PNGs in my case since they seem to work best for screen shots) to propose the DPI setting it will use on import. So, if I can be bothered I can change the DPI "setting" in Photoshop (without affecting the image in any way) and FM will then propose importing at 96DPI. In practice i don't bother and it's now second nature to set FM to import at 96DPI.

Rob The picture is only there to keep the sound in sync

Alan Roberts
Alan Roberts's picture
Offline
Joined: May 3 1999

Incidentally, my Dell (claims to be 120dpi) is 1920pixels and 331.5mm wide, so the actual dpi figure is 1920/331.5*25.4=147.113.... The DVC special (claims to be 96dpi) is 1920pixels and 367.3mm wide, so the actual dpi figure is 1920/337.3*25.4=132.774dpi. Neither is close to either 96 nor 120, but what the heck...

I'm happy with this situation, because it gives me both the commonly used "resolutions", and I have to be able to check how my software works at differing resolutions.

Ho um :)

Get my test cards document, and cards for 625, 525, 720 and 1080. Thanks to Gavin Gration for hosting them.
Camera settings documents are held by Daniel Browning and at the EBU
My book, 'Circles of Confusion' is available here.
Also EBU Tech.3335 tells how to test cameras, and R.118 tells how to use the results.