Getting desparate - Image Sizing

[Anewman]

Assuming that the upsamplers offer comparable quality (as I suspect they do), you gain nothing by doing it yourself.

The only thing I feel that I gain, is FINAL PREVIEW and more adequate wysiwyg type of workflow. I apply the final Sharpening b4 uploading and have the lab turn off all auto adjustments.

Obviously I agree that if one is not 100% sure what DPI the lab uses it could end up being worse, That is why I would not reccomend resizing unless one knows EXACTLY what DPI the machine uses.

 

[Amy]

Here is the best explination I can give.

each camera takes a picture, that picture is made up of lots of pixels. Each pixel is a rectangular box that is of a certain color. Because they are small and there are lots of them, you get smoth lines, curves and colors in your picture.

Your camera typically will take a picture with a ratio, most are 4x6 which means you will have 6 pixels horizontally to every 4 pixels veritcally when the final product of pixels is assembled.

If your printing a 4inch x 6 inch photo, or anything with that ratio, this is a good thing, so it would print 4x6, 8x12, 16x24, all are the same ratio.

If you wish to change the ratio of the print you have 2 options.

Option 1, crop out the part of the picture that conforms to the new ratio. so if you wanted a 5x7 picture So in your 4x6 size, you would have 4x5.6 ratio. You would therefore have to cut off .4 of your picture to get it to fit.

Option 2 resize and not crop. This would in the example above squeeze the .4 of the picture that would be cut off in the option above into the 5.6 availibe space, but that makes the pixels become narrower than they were originally therefore distorting the picture.


Now if your camera MP would normally print a good 8x10 picture, but you needed an 80x100 picture, you could blow it up, but what that does is just increase the size of the individual pixel, so the little rectanglar box becomes a much larger rectangular box, there by making your picture not as good, because the eye can now see the little boxes rather than being fooled into thinking it is a smoth line and curves.

Does that make any sence or did I just confuse you more?

No, you didn't confuse me; that sounds like the way my brain was headed. Life was easier before pixels, but definitely not as much fun!

Thanks, Master Mason!

 

[extreme8]

Why? Because you trust your software's ability to upsample more than theirs?

Absolutely.

Costco's photo company "Dry Creek Photo" reccommends;
If the lab can not tell you what resolution their printer uses, the safest course is to use the 300 ppi values in the table below for your first print order.
This is an excerpt from a longer document which goes on to explain how you can use the legend on the back of the print to find out if their printer is running at 300 or 320 dpi before preparing your next order.
Since the OP specifically mentioned printing at Costco I figured it was applicable, so I included the info in my post.

You can't create picture information where none existed before.

It is impossible to create data where none exists, but some programs are better than others when guessing what is "in between" the existing pixels.
I trust the techniques I've developed using photoshop CS2, I know how it's interpolation works with different resizing methods, and I know what results I can expect.
If making an extreme enlargement I use programs that take advantage of fractal based algorithms and they do a phenominal job.

I have no doubt that Costco RIP's can usually do a fine job of printing Aunt Sally's 75 dpi pictures at 8x10.
On the other hand I got an 8x10 done at Walgreens (I needed it in a hurry) that was sent to them at the wrong size/resolution and it was in the garbage can before I made it back to the car.
As they say on TV, your mileage may vary.

I like to eliminate as many variables as possible.
My monitor is calibrated and I normally only print places that supply a printer profile for each paper I plan to use. I soft proof before making final adjustments and I apply the profile before it's saved. I send the image at exactly the size and resolution that the output device wants. I upload at full size and turn off auto correction.
According to the legend on the back of the Costco prints I'm within .03% of optimal size.

I believe that the more decisions you can remove from the automated equipment the better your end result will be.

OK, gonna stop now.

It feels like this is getting a bit tense and there's no need - we've got too many common interests to get sidetracked by - of all things - resolution and image sizes.
I Really like you guys (and girls!) and the atmosphere here and don't feel that it's worth getting our backs up over something as silly as workflow.
See y'all in the weekly contest thread!

 

[Groucho]

Don't worry about me, I'm not getting tense, and I'm sorry if I came off that way or said something to make you tense. I'm just curious if I'm missing something. This is interesting on an academic level.

I understand the idea that there are different techniques for resampling (I mentioned recently that Irfanview has six built-in and there are, I'm sure, a handful of plugins for Photoshop that claim to do a better job than Photoshop) - but again, no matter how good they are - you can't take a photo, double its resolution, and have it look as good as a photo that started out at the higher resolution. (Similarly - don't expect that "HD upscaling" DVD player Best Buy wants you to buy to make your DVDs look like HD. It won't.) I would hope that a decent photo lab will use software that does a decent job of sizing your prints. Heck, I have a photo that started out as a mere 1280x1024 that's printed at something like 22x17 (it was a freebie done to demo a $100k Canon printer - thanks Canon!) and it's gorgeous. Up close you can certainly see the pixels but at a normal distance, it looks pretty good. I also had a 5mp print made. I'd be really shocked if resampling ahead of time changed the final image in any way.

My feelings (and I won't be offended if you disagree) at that the differences between doing it yourself and having them do it (assuming that they're not total dolts) are so minor (if they exist) that you'll be probably have to pull out a magnifying glass to see them - we are talking 300+ dpi here. And, not to take anything away from the intention - but this is a photo to give out to a kid's sports team, it's not something that will be hung up at an exhibition. I think worrying about resampling makes things more complicated than they need to be.

Dry Creek Photo's recommendation is certainly a safe one but they're obviously marketing more towards the photo professional, who is probably working with very large files in Photoshop, adding text and graphics, retouching fine detail, etc. In other words, something other than a photo straight from the camera. By the way, I did a quick search on them to see what they have to say and there are many complaints about them taking money for work that is never done - something to be careful about. Not that this makes the advice any better or worse.

Probably the best bet is to ask the lab not only what dpi their printer uses, but what software they use to prepare the prints (possibly proprietary to that printer? I'm not familiar with what goes on behind the doors at the labs.) If you can find out how that software works, you'll have an idea what to expect from it.

 

[jann1033]

ok maybe i totally lost this info somewhere along the way but

if you are editing your photos and don't really know what size you are going to end up printing it( if at all) what would be a safe size to leave the resolution and size if i crop it ( missing manual book really seems to recommend using the rectangular marquee tool insteaad of the crop in PSpE5 as the crop automaticaly resamples) that would give me the most leeway to print a poster or a 8x10 or whatever later? or does it matter what i do ? just tell me numbers and i will fill them in the blanks no questions asked
thanks

 

[Anewman]

ok maybe i totally lost this info somewhere along the way but

if you are editing your photos and don't really know what size you are going to end up printing it( if at all) what would be a safe size to leave the resolution and size if i crop it ( missing manual book really seems to recommend using the rectangular marquee tool insteaad of the crop in PSpE5 as the crop automaticaly resamples) that would give me the most leeway to print a poster or a 8x10 or whatever later? or does it matter what i do ? just tell me numbers and i will fill them in the blanks no questions asked
thanks

The safest bet would be to leave them at the native(full) resolution, IMO there would be no need to resize or crop until one is sure of desired output.

And no the crop tool will not automaticaly resample unless you put in a value in the DPI(resolution) box, leave it blank and it will not resample. With the rectangle marquee, you can not control the aspect ratio of your crops. At least that is how it is supposed to work.

 

[extreme8]

...differences between doing it yourself and having them do it (assuming that they're not total dolts)...

There's the sticking point - I'm not prepared to make the assumption that the people you deal with at the local minilab aren't total dolts. Although I realize there are some very competent people working photo labs, I've seen some pretty nasty work come out of mass marketers like Walmart, Walgreens & CVS.
I might be a complete dolt myself, but at least I know that ahead of time and don't have any expectations otherwise.

I think worrying about resampling makes things more complicated than they need to be.

I'll have to agree.
I guess most photo labs just require a certain size and don't bother even mentioning resolution, so I'm sure they do a job that the average shooter is happy with.

I just did a search for Dry Creek Photo and saw the same negative comments you did.
I was somehow under the impression they were running the photo centers at Costco - evidently they just provide the icc profile service. Interesting.
I do agree that their advice is still sound.

 

[jann1033]

not to hijack Jen's thread but maybe she wants to know( i hope hehe)

when i first open a newly downloaded picture it's always like 2x3 ft( roughly) and at 72 dpi. when you say leave it at native resolution( guessing you mean size as well) even if i crop it because say I only want part of it as the picture, i should leave the resolution at 72? the reason I'm asking is i edited one picture i wanted to print and ended up only able to print it at 5x7 when i wanted larger and not sure how i got to that point but would rather not have it happen again...was that due to how much i cropped before i changed the resolution? should i change the resolution before i crop or does it matter...ie if i crop first will it still have 200dpi if it would have had 200 dpi if i crop it after i change it...
am i making any sense to anyone????????????????????

i guess my thinking is if i change it pre crop it has more pixels to work with to pack together but if i change it post crop it has to stretch what's there so may not be able to get as good of resolution for the size....but since that would make sense( to me) it's probably not the case

 

[PoohJen]

Hijack away; y'all lost me at "Hello".

 

[Groucho]

Hey, the dpi thing is just like in the other thread.

DPI is a measure of density and is meaningless in the digital world - it only applies when you're dealing with prints. (see footnote) When the picture is on your PC, the only thing that matters is the pixels. The JPG does store dpi information, but it's only used when printing. It can also be changed at any time without affecting the picture itself.

Furthermore, the dpi information is usually ignored when printing - usually the picture is sized upwards or downwards to match what size you want it printed at - 4x6 or whatever.

IMHO there are only a few cases where the DPI information is important... such as:
Where you've scanned something at a specific DPI and want to print it out again at exactly the same size
If you're printing something with measurements (like inch markers or similar)
If you're created a poster or collage or something, where you're adding different photos, adding text, etc - knowing the DPI will let you create a proper "canvas size" ahead of time without worrying about losing any quality by making an original image too small.

If you have printing software that will only let you print at the dpi listed in the jpg - time for better printing software! If you're using special paper like ones with perforated edges to pop out a couple 4x6 prints from an 8.5x11 sheet, then you need to use something that supports them. If you're printing to a regular sheet, Irfanview can print a picture at any size you want - original DPI, shrunk, stretched, centered, whatever. Most photo programs should have such functionality.

To sum up: just ignore the dpi settings when doing things like cropping, retouching, etc. Keep the photo at original resolution (not bigger or smaller) and you'll be fine.

extreme8: I do agree that one should never underestimate the ineptitude of the work done at some places. But hopefully even with a dolt, the software is doing the work and as long as it's not junk software, the results should be good. I would guess that the dolt-damage would be likely to manifest itself in poor brightness and contrast rather than in resizing artifacts... but who knows?


(footnote from above: This may not be true forever, Microsoft is attempting to get some kind of standard DPI measurements into future versions of Windows, so that you can run a monitor at any resolution and you'll see the same size fonts, windows, pictures, etc - this will be done via 3D acceleration, with the Aero interface. I ran Vista for a while but never on a system that could use the Aero interface so I'm not sure how functional that is in Vista. I suspect that it'll be a bit iffy for the moment, but in the future it may work acceptably and be replicated in other operating systems. SOMEthing needs to be done - these new laptops with 1920x1200 screens are almost unusable at the native resolution due to the pixels being so tiny.)