I'm having some 12x18 prints made at White House Custom Color and I have a few questions related to upsizing/uprezing the files for print. The prints will be coming from a Nikon D80 10.2 MP (8.6x12.9 inch native size) or D7000 16.2 MP (10.9x16.4 inch native size).
My question is...is it better to handle the upsizing myself using photoshop (I'm using CS4) or just send the file to WHCC at it's native size and allow their printer to upsize the file?
If you agree with upsizing the file in PS before sending it to the print lab, how do you go about resizing the image? I feel the best way to resize in PS is to use the "image size" dialog and change the document size to the desired size. Do you keep the resolution at 300ppi when upsizing? What do you do with the scale styles, constrain proportions, resample image check boxes? Do you sharpen the image before or after resizing?
Hope this all makes sense. Thanks in advance for any help/advice you can provide.
When I first started using WHCC I sent in some 8x10 experiments, including two comparing exactly this with my D80. Granted that two is a very limited sample, but to my eye the image I upsized to 300dpi and then sharpened was a tad sharper (better) than the image I sharpened at native resolution and then had WHCC upsize.
I upsize to 300dpi. I use bicubic smoother, check constrain proportions and resample image, and then (on my monitor, others may vary) sharpen with the image viewed at 50% in Photoshop.
Scott, in the printing DVD set that Jeff Schewe and Michael Reichmann did, Jeff stated that it's best to upres a full 200% which forces Photoshop to do some algorithm that is best for image quality. Then sharpen it and reduce the image down to where you need it to go. Reducing it down after sharpening is one of the steps in the Marc Adamus Web Sharpening procedure that has worked extremely well, so maybe this is the way to go for your print. I haven't tried it yet but it makes sense, if you feeling kind of guinea pig-like lately.
You might want to try this technique which I've found to be very effective for upsizing:
1. Open the Image Size dialog box
2. Type in the dimension you want as your final print size
3. Adjust the resolution upward to 360
4. Choose Bicubic Sharper (yes, sharper) as the resampling method
5. Click "OK"
This method has been in the last few Scott Kelby books which he said he got from Vincent Versace. I used this to make a few larger prints (24x30) that I sold and they looked very good.
Scott - I would uprez myself for maximum quality and control. Here's a tutorial that Nat Coalson posted over at RMNP a few years back when he making a 40x60 from an 8.3MP file. It addresses most of your questions and, even if you don't have the programs that Nat references, describes the general workflow for significant upsizing. Nat's an Adobe certified lightroom and photoshop expert, and an excellent printer. If you wanted to shoot him some questions, I'm sure he'd be happy to help; just search for "Nat" on the forum to get his email link. Also, posting your question on the RMMP forum would be a good move - there's a ton of printing expertise over there.
Nat Coalson:
I was able to make the enlargement to 40x60 inches from an 8.3 MP capture using the following method. (Actually, I cropped the image in Lightroom, so I started with a file with an even smaller native resolution of 3128x2085.)
In Lightroom
1. Process the DNG file with no sharpening. Using the controls in the Develop module, get the file as "perfect" as possible with regard to cropping, color and contrast. Apply very slight noise reduction if necessary, and make sure to correct any chromatic abberation, which dramatically affects sharpness.
2. Export from LR as 16-bit ProPhoto PSD (or TIF, if you prefer) at 240 ppi with no resizing.
In Photoshop
3. Before doing anything else, apply PhotoKit Capture Sharpen Expert > Digital Hi-Res Sharpen.
4. Make any necessary Adjustments that were not taken care of in Lightroom, such as localized dodging and burning or selective color adjustments.
5. Apply PhotoKit Creative Sharpener > Sharpening Effects > Luminance Sharpen 3 (this selection will depend on the characteristics of the image you're working with; it's a good idea to experiment to learn the best kind of sharpening for different types of images).
6. Save out the layered Master PSD file, then flatten it. (And don't save again just yet.)
7. In Image Size, upsample the image 200% using Bicubic Smoother.
8. Zoom in to see if there was any apparent loss of detail from the resampling. If so, apply another round of sharpening using either PhotoKit, Unsharp Mask or Smart Sharpen, depending on the needs of the image.
9. Upsample the image to its final size, again using Bicubic Smoother. Apply another round of light sharpening if necessary.
10. Flatten again and Save a copy of the resized file for printing.
11. Before printing, apply PhotoKit Output Sharpener > Inkjet > 240 ppi glossy
In summary: making good enlargements from digital captures requires several rounds of gentle sharpening interspersed with several incremental rounds of upsampling. You will get much better results - and have more control over the final image - if you don't try to do all your sharpening or resampling in one step. And of course, some images will have more potential for huge enlargements than others, and no amount of sharpening will make up for an out-of-focus image. (5D owners properly using these techniques may be amazed by the possibilities for enlargement.)
I strongly recommend that you work on your Master File at native resolution, as far through the pipeline as possible. Do your capture sharpen, creative sharpening and all adjustments on the Master File and when you are done with all your edits, do your upsampling on a flattened copy of the file. This way, you can always go back to your Master file and make new derivatives for specific print sizes without having to do your primary work on huge files. And keep in mind that the ideal settings for any sharpening depend on the pixel resolution of the file.
Last year some of these techniques were featured in a very interesting article in Digital Photo Pro, written by Jeff Schewe. Jeff is one of the founders of PixelGenius, who makes PhotoKit. PK Sharpener was developed by the late digital guru Bruce Fraser.... and nobody in the business knew more about digital imaging than Bruce. In my opinion, spending $99 to add PhotoKit Sharpener to Photoshop is one of the best ways to dramatically improve your images. (Note that the CS3 version of PK is still in Beta; for this reason I'm still mainly using CS2.) http://www.pixelgenius.com
I hope readers find this information useful and I will be very interested to hear some "enlargement success stories" from others... and let me know if you need any big prints from my 9800!
Thanks to everyone for all of the suggestions. I'm amazed by how many ways there are to achieve the same results. After researching this extensively, I decided to use Floris van Breugel's technique for upsizing/uprezing images for print as outlined HERE. Thank you Floris if you happen to read this post. When I get my first print back from WHCC I will report back on how it turned out.
I sometimes find this subject almost humorous. There are so many opinions and they often differ so much! This is a subject I have devoted an incredible amount of time, research and practical testing on (and still am). I have dialogued with the Adobe Gurus/engineers and have done boat loads of real world testing. This is what I do full time for high profile (and lower profile) fine art landscape gallery clients (and my own work). I have to know my stuff (and I want to because of my insatiable appetite to master prints). It is too complicated to explain in length here but if you look me up on my website and give me a call I would gladly give you a little free consulting.
I can say this:
1. At small sizes 18 inches or less the differences are likely going to be unnoticeable.
2. IMO (and a strong one at that) Genuine Fractals being the best uprezzing software is a myth based on faulty testing. It keeps getting perpetuated because of apparent dramatic visual comparisons that don't factor in the extra sharpening a "Smoother " file can take.
3. Nothing yet beats Advanced use of PS USM in terms of sheer quality and control.
4. It is often beneficial to use a combination of High Pass sharpening, USM and even a touch of grain simulation. So, the best fundamentals are: 1. Capture Sharpen in ACR. 2. Uprez using Smoother. 3. High Pass to tighten up the image a bit. 4. USM with advanced methods (or USM then High Pass). 5. Grain Sim.
5. Third party software will not size or sharpen better than advanced custom sharpening and Bicubic Smoother upsizing (although third party software is almost always much less convoluted - a major benefit to many).
6. Capture sharpen in ACR (very carefully based on the commonly known rules there). If this is done carefully and appropriately it will allow for a small edge in print quality over not doing it. I know a secret little tip here that can extract a little more detail out of your raw file...
7. Always sharpen AFTER upsizing (there is a difference). If you sharpen then size you will exaggerate the unpleasant halos into visible acuity.
8. Use Bicubic SMOOTHER (not sharper). I used to believe the Bicubic Sharper myth (was originated from a popular Scott Kelby PS book) . Basically "Sharper" adds an uncontrollable amount of sharpening (edge halos) into the upsize. It is much better to completely control your sharpening (where you want it and how much you want it). Nothing on the planet yet beats Bicubic SMOOTHER. Also, the Genuine Fractiles myth is based on the edge detection sharpening that is built into the algorithm. But in the end can not take as much sharpening after uprezing. Bicubic Smoother can handle more aggressive sharpening after the uprez, with less artifacting and a more natural appearance.
9. Use max ppi resolution on small prints (like 400 or 360 or 300ppi...) but on huge enlargements you can usually drop that down beneficially to around 200ppi (depending on the printer).
10. If in 16 bit, go to the LAB L channel for USM (it can slightly beat out RGB Luminosity mode sometimes). If the file has suffered some abuses then RGB Luminosity mode may be the better option. If the file is a robust data file carefully worked then the color space jumping will not effect quality.
11. *If you understand the Layer Style "Blend If" sliders you can control sharpening and grain fall off (for shadows and highlights) easily and very powerfully.
12. For very fine detail images often having the USM settings to 500 Amount, Threshold ALWAYS on 0, and then easing up the radius slowly works great. Maybe ease it up one notch (or point) higher than looks perfect on your monitor. That should give you a very good start.
13. Use a combo of High Pass and USM. I break what seems like convention and often do the High Pass first for a general, very moderate tightening up of the images, before USM. I also do the High Pass but then apply the finished effect in Luminosity mode so to avoid any slight color shifting/boosting. You can not do a High Pass in the LAB L Channel.
14. For high res images 300+ ppi DON'T view at 50%. View at 25%. 50% works best for lower res images (like around 200ppi). DONT view at odd sizes. Just 25% or 50% depending on res.
15. You can make an edge mask (adding to the whole mix after the uprez) to protect smoother areas (from the USM) and adjust its opacity. I have found this little less important these days when a decent Capture Sharpening is done in ACR due to the masking available there. I think if the ACR Capture Sharpening was not so good and the masking there not available, I would rely on edge masks more. Basically the ACR Capture Sharpen enhances detail (when done right) in all the areas except smooth surfaces and shadow areas. Then once you are uprezzed those areas have already had some built in protection. There is Capture Sharpening and there is also something sometimes referred to as anti-convolution. If you choose your settings in ACR Capture Sharpening well I believe you can even extract a very small amount of actual detail from your raw file otherwise impossible.
16. Choose the newest rendering intent in ACR or Lightroom (which has a lot more detail than the old rendering). The new rendering is competitive (in terms of detail) with any raw converter out there.
17. Make sure your monitor is set at its highest resolution viewing.
18. Sharpen just a tad too crispy, since prints are naturally quite a bit softer than the monitor (squares are turning into circles). I suggest over-sharpening a little bit vs under-sharpening. You may be surprised how soft a print is in comparison.
19. If used carefully and maybe at or just under the limits of human acuity GRAIN SIMULATION can both camouflage digital artifacting caused by sharpening and even add the illusion of MORE detail (and I'm not talking about a grainy looking photo)! If mastered this will make image to look more detailed and give the image a more organic look.
20. ACR's somewhat new grain simulation engine does a great job of producing natural, almost organic like grain effects. I do it on a sharpened layer and ease the effect into the sharpened image at different opacities in different areas, as a very last step.
21. Maybe check out the Adobe Forums where the minutia of these issues are regularly argued and tested rigorously!
22. Read all of Bruce Fraser and Jeff Schewe's work then test, test, test!
Floris's write up is pretty good. If you combined that with some of the above suggestions you will certainly be on your way to mastering sharpening.
Good luck.
Posted by Scott Smorra on June 12, 2011 at 5:20 PM
Wow Mark, thanks for taking the time to write up these tips on uprezing/sharpening. I truly appreciate it! You have given me lots of things to think about.
For capture sharpening in ACR/Lightroom, I'm using Bruce Fraser/Jeff Schewe's technique as outlined in their Real World Sharpening Book. It seems to work very well for me. One question for you...is there a maximum for the "amount slider" which should be applied in the capture sharpening phase? I've generally been using an "amount" of 50-60 for landscape images and everything seems fine, but I wondered if you have an opinion on this.
How aggressive are you with USM when you apply the USM after high pass sharpening? I ask because I seem to be getting pretty good results with just high pass sharpening with some masking of areas I don't want to be over sharpened.
Thanks for the tips on the layer styles and using ACRs grain simulation and viewing at 25% for smaller prints.
Once again, thanks for taking the time to write this all up! You are obviously a professional and have a great understanding of the tools required to make outstanding fine art prints.
In ACR view at 100% and capture sharpen just up to the point of no visible ill effects. Tighten up the image but see no artifacting or crispiness or anything. If in doubt, back off a tad. You can look very close on your monitor though. And a good monitor helps. The little secret (and it may be controversial) is to take your radius all the way to the left (lowest setting) and detail all the way to the right (highest setting) mask moderately (or maybe almost not at all if the image is completely full of high frequency details) then ease up the "amount". With the radius all the way left and detail all the way right, from what a adobe top dog guru told me (and I see in practice and test prints) is that it actually may extract a very small tad of ACTUAL detail out of the raw file otherwise not accessible. I have tried all kinds of settings and this is my preferred. Personally I often bring in multi versions of capture sharpening into PS and then manually erase things like the sky... As far as actual amounts... it really depends on the f/stops and lenses you use (the sharpness of your file). With primes and killer technique I have had images not be able to handle more than about 20-30 amount. On softer stuff upwards to 50-60 and maybe as high as 80 rarely.
USM is not very aggressive because High Pass has done about 1/2 the work (and you can try vise versa with high pass last too). Again just look at the image at the right viewing distance and get it looking perfect, then go just a bit beyond that (a tad crispy). Then do test prints and you will eventually get a very good grip on it. I do find it amazing how bad some gallery prints are these days. Mastering sharpening is really key! Good luck.
As far as grain. Don't underestimate its help. Also I never add it to the raw file. It is a very last step in workflow.
Discussion
9 repliesMy question is...is it better to handle the upsizing myself using photoshop (I'm using CS4) or just send the file to WHCC at it's native size and allow their printer to upsize the file?
If you agree with upsizing the file in PS before sending it to the print lab, how do you go about resizing the image? I feel the best way to resize in PS is to use the "image size" dialog and change the document size to the desired size. Do you keep the resolution at 300ppi when upsizing? What do you do with the scale styles, constrain proportions, resample image check boxes? Do you sharpen the image before or after resizing?
Hope this all makes sense. Thanks in advance for any help/advice you can provide.
I upsize to 300dpi. I use bicubic smoother, check constrain proportions and resample image, and then (on my monitor, others may vary) sharpen with the image viewed at 50% in Photoshop.
1. Open the Image Size dialog box
2. Type in the dimension you want as your final print size
3. Adjust the resolution upward to 360
4. Choose Bicubic Sharper (yes, sharper) as the resampling method
5. Click "OK"
This method has been in the last few Scott Kelby books which he said he got from Vincent Versace. I used this to make a few larger prints (24x30) that I sold and they looked very good.
Nat Coalson:
I can say this:
1. At small sizes 18 inches or less the differences are likely going to be unnoticeable.
2. IMO (and a strong one at that) Genuine Fractals being the best uprezzing software is a myth based on faulty testing. It keeps getting perpetuated because of apparent dramatic visual comparisons that don't factor in the extra sharpening a "Smoother " file can take.
3. Nothing yet beats Advanced use of PS USM in terms of sheer quality and control.
4. It is often beneficial to use a combination of High Pass sharpening, USM and even a touch of grain simulation. So, the best fundamentals are: 1. Capture Sharpen in ACR. 2. Uprez using Smoother. 3. High Pass to tighten up the image a bit. 4. USM with advanced methods (or USM then High Pass). 5. Grain Sim.
5. Third party software will not size or sharpen better than advanced custom sharpening and Bicubic Smoother upsizing (although third party software is almost always much less convoluted - a major benefit to many).
6. Capture sharpen in ACR (very carefully based on the commonly known rules there). If this is done carefully and appropriately it will allow for a small edge in print quality over not doing it. I know a secret little tip here that can extract a little more detail out of your raw file...
7. Always sharpen AFTER upsizing (there is a difference). If you sharpen then size you will exaggerate the unpleasant halos into visible acuity.
8. Use Bicubic SMOOTHER (not sharper). I used to believe the Bicubic Sharper myth (was originated from a popular Scott Kelby PS book) . Basically "Sharper" adds an uncontrollable amount of sharpening (edge halos) into the upsize. It is much better to completely control your sharpening (where you want it and how much you want it). Nothing on the planet yet beats Bicubic SMOOTHER. Also, the Genuine Fractiles myth is based on the edge detection sharpening that is built into the algorithm. But in the end can not take as much sharpening after uprezing. Bicubic Smoother can handle more aggressive sharpening after the uprez, with less artifacting and a more natural appearance.
9. Use max ppi resolution on small prints (like 400 or 360 or 300ppi...) but on huge enlargements you can usually drop that down beneficially to around 200ppi (depending on the printer).
10. If in 16 bit, go to the LAB L channel for USM (it can slightly beat out RGB Luminosity mode sometimes). If the file has suffered some abuses then RGB Luminosity mode may be the better option. If the file is a robust data file carefully worked then the color space jumping will not effect quality.
11. *If you understand the Layer Style "Blend If" sliders you can control sharpening and grain fall off (for shadows and highlights) easily and very powerfully.
12. For very fine detail images often having the USM settings to 500 Amount, Threshold ALWAYS on 0, and then easing up the radius slowly works great. Maybe ease it up one notch (or point) higher than looks perfect on your monitor. That should give you a very good start.
13. Use a combo of High Pass and USM. I break what seems like convention and often do the High Pass first for a general, very moderate tightening up of the images, before USM. I also do the High Pass but then apply the finished effect in Luminosity mode so to avoid any slight color shifting/boosting. You can not do a High Pass in the LAB L Channel.
14. For high res images 300+ ppi DON'T view at 50%. View at 25%. 50% works best for lower res images (like around 200ppi). DONT view at odd sizes. Just 25% or 50% depending on res.
15. You can make an edge mask (adding to the whole mix after the uprez) to protect smoother areas (from the USM) and adjust its opacity. I have found this little less important these days when a decent Capture Sharpening is done in ACR due to the masking available there. I think if the ACR Capture Sharpening was not so good and the masking there not available, I would rely on edge masks more. Basically the ACR Capture Sharpen enhances detail (when done right) in all the areas except smooth surfaces and shadow areas. Then once you are uprezzed those areas have already had some built in protection. There is Capture Sharpening and there is also something sometimes referred to as anti-convolution. If you choose your settings in ACR Capture Sharpening well I believe you can even extract a very small amount of actual detail from your raw file otherwise impossible.
16. Choose the newest rendering intent in ACR or Lightroom (which has a lot more detail than the old rendering). The new rendering is competitive (in terms of detail) with any raw converter out there.
17. Make sure your monitor is set at its highest resolution viewing.
18. Sharpen just a tad too crispy, since prints are naturally quite a bit softer than the monitor (squares are turning into circles). I suggest over-sharpening a little bit vs under-sharpening. You may be surprised how soft a print is in comparison.
19. If used carefully and maybe at or just under the limits of human acuity GRAIN SIMULATION can both camouflage digital artifacting caused by sharpening and even add the illusion of MORE detail (and I'm not talking about a grainy looking photo)! If mastered this will make image to look more detailed and give the image a more organic look.
20. ACR's somewhat new grain simulation engine does a great job of producing natural, almost organic like grain effects. I do it on a sharpened layer and ease the effect into the sharpened image at different opacities in different areas, as a very last step.
21. Maybe check out the Adobe Forums where the minutia of these issues are regularly argued and tested rigorously!
22. Read all of Bruce Fraser and Jeff Schewe's work then test, test, test!
23. HAVE FUN!
Good luck.
For capture sharpening in ACR/Lightroom, I'm using Bruce Fraser/Jeff Schewe's technique as outlined in their Real World Sharpening Book. It seems to work very well for me. One question for you...is there a maximum for the "amount slider" which should be applied in the capture sharpening phase? I've generally been using an "amount" of 50-60 for landscape images and everything seems fine, but I wondered if you have an opinion on this.
How aggressive are you with USM when you apply the USM after high pass sharpening? I ask because I seem to be getting pretty good results with just high pass sharpening with some masking of areas I don't want to be over sharpened.
Thanks for the tips on the layer styles and using ACRs grain simulation and viewing at 25% for smaller prints.
Once again, thanks for taking the time to write this all up! You are obviously a professional and have a great understanding of the tools required to make outstanding fine art prints.
In ACR view at 100% and capture sharpen just up to the point of no visible ill effects. Tighten up the image but see no artifacting or crispiness or anything. If in doubt, back off a tad. You can look very close on your monitor though. And a good monitor helps. The little secret (and it may be controversial) is to take your radius all the way to the left (lowest setting) and detail all the way to the right (highest setting) mask moderately (or maybe almost not at all if the image is completely full of high frequency details) then ease up the "amount". With the radius all the way left and detail all the way right, from what a adobe top dog guru told me (and I see in practice and test prints) is that it actually may extract a very small tad of ACTUAL detail out of the raw file otherwise not accessible. I have tried all kinds of settings and this is my preferred. Personally I often bring in multi versions of capture sharpening into PS and then manually erase things like the sky... As far as actual amounts... it really depends on the f/stops and lenses you use (the sharpness of your file). With primes and killer technique I have had images not be able to handle more than about 20-30 amount. On softer stuff upwards to 50-60 and maybe as high as 80 rarely.
USM is not very aggressive because High Pass has done about 1/2 the work (and you can try vise versa with high pass last too). Again just look at the image at the right viewing distance and get it looking perfect, then go just a bit beyond that (a tad crispy). Then do test prints and you will eventually get a very good grip on it. I do find it amazing how bad some gallery prints are these days. Mastering sharpening is really key! Good luck.
As far as grain. Don't underestimate its help. Also I never add it to the raw file. It is a very last step in workflow.