Post Crop Vingette in LR (2.5) Creates Banding in JPEGs for Lab

[Theodore]

Hello all -

I sent some files off for printing with Millers/MPIXpro and one 8 x 10 that came back really surprised me.  It's a photo of two girls in the beach on an overcast day (D700, Nikkor 85 1.4 AF-D).  In LR 2.5 I had applied a post crop vingette and didn't notice an issue in the viewing window of LR.  I saved the file as an 8-Bit sRGB per the usual commercial lab requirements and shipped it off electronically for printing.  What came back looked like a beverage had been spilled in the middle of it with gradient / color banding working our from the middle following the vingette pattern.  

I was surprised and this and for a moment thought it might be a lab error but then and opened the .jpg file and saw the same banding (although it's more pronounced when printed).  I searched the ADOBE forums and found a topic on this back from LR 2.1 where someone noted that the problem was the LR didn't dither when it saved images to 8 Bit JPEGs, but that PS would dither.  Now, I haven't seen this before but then again, I didn't see this in LR either - only after seeing it so clearly in print and then looking at the JPEG exported file did I see it.

My question is, is there some setting in LR that I'm missing that will address this issue (for example if it is a dithering issue, is there a setting to have the dithering done when exporting an image with a vingnette to JPEG.?) Or is this an issue yet to be addressed in LR?  

Here's the LR 2.5 file both with and without the -16 post crop vingette (attached).

Thanks everyone,
Theodore

[Theodore]

Thanks for the response John.  I'll have to think through the point that you made re: a point curve.  I have to admit to needing to study up to fully understand how to do that.  Of course, what I hope for is a software fix that simple dithers the effect better for JPEG printing output so that the vingnette sliders are reliably useful (i.e. I can use it and not have to worry that it will look awful when it goes to the lab - right now vingnette and direct lab-to-customer fulfillment from my site don't appear to be compatible and that's a real issue).  If anyone is aware of an Adobe software fix in the works as a LR improvement, that would be great to hear about.

Yes, it's a shame that commercial printing is so limited in terms of the files it can reproduce, where software, cameras, and inkjet printing can create images in the larger color spaces / greater bit depth.  I'd love to be able to print myself or have West Coast Imaging do the Epson prints, but it's a matter of time and expense and a WHCC, or Millers, or Bay Photo or... are well suited to portrait and event needs based on the costs, i.e. it's better to have a $2 8x10 when you're selling for $20 than a $10 Epson 8x10 and to have the lab do it when someone is ordering a 350 image proof set.  

Thanks again for the response John.  It was certainly interesting to hear about your encountering this on your large print and how you've addressed it.

[Panopeeper]

(This is not mitigate the least what John wrote.)

Have you recorded the raw data with lossy compression?

[Theodore]

(This is not mitigate the least what John wrote.)

Have you recorded the raw data with lossy compression?

Hi Pano, no, just a straight uncompressed RAW file into LR.  

It's worth noting that I've seen many post-crop vingette files that are fine and print fine when output to a JPEG for the lab, so it certainly appears to be an issue that accentuated by the lighting involved.  On the Adobe forum, I found a string on this and it was noted that LR doesn't dither the applied vingnette on conversion to a JPEG and so as an 8 Bit JPEG it shows up as banding in certain lighting scenarios.  The forum noted that PS could save such a file with the dithering, but (and while I haven't tried it in PS CS4) that doesn't address the issue in LR, which I'd hoped to use the vingette effect with confidence on portrait and event photos that are ultimately going to be printed at a professional lab (in this case Millers, but as John noted, he got the same result on an image with WHCC on a 2x3).

[leuallen]

Just a quick thought. Since Photoshop dithers and might handle the vignette properly, why not apply the vignette in LR, then open in Photoshop with LR settings, then do a save as from Photoshop. See if the Photoshop saved file prints better than the LR one. This is assuming that LR transfers the vignette with 16 bits. Might work. Don't have a file to try it on. The extra step doesn't seem like it would take much effort or time.

Had the problem once. Don't remember what I did though. Anyway got past it whether it was to do it in Photoshop or trash the vignette idea entirely.

Good luck,
Larry

[Panopeeper]

just a straight uncompressed RAW file into LR

I asked the "lossless" question, because it does occur, that the lossiness causes banding.

This now has nothing to do with the original topic, but why don't you use the losslessly compressed format? It is fully equivalent to the uncompressed in content, but much smaller. This was an issue with some earlier Nikon models, that the compressed version was always lossy, but now there are six formats to choose from (12bit or 14bit, and uncompressed, losslessly compressed or lossily compressed).

[Theodore]

I asked the "lossless" question, because it does occur, that the lossiness causes banding.

This now has nothing to do with the original topic, but why don't you use the losslessly compressed format? It is fully equivalent to the uncompressed in content, but much smaller. This was an issue with some earlier Nikon models, that the compressed version was always lossy, but now there are six formats to choose from (12bit or 14bit, and uncompressed, losslessly compressed or lossily compressed).

Hi Pano - you know I went and checked what I thought I knew, and I was wrong (although I did set it to lossless over a year ago, so hopefully my memory isn't so bad) and it is losslessly compressed.

[Wayne Fox]

These labs are all using Durst Lambdas for printing. While very high end machines and good quality printing comes from them, they are perfect in that they print any banding or hard gradients (that are barely evident in a file) very noticeably. Whereas, a print on an Epson doesn't show these things anywhere as much.

While I believe they might have a Lamda or two,  Millers and WHCC use mostly a combination of Durst Theta and Noritsu printers. Millers facility is quite amazing, I believe I counted 20 of these printers in all lined up when I toured their main plant a few years ago.    The Durst Theta uses a LED/ fiber optics system as the imaging source instead of a laser which is what the lamda and most of the Noritsu's use.  I believe both labs also have a Chromira for extremely large prints, which also uses LED technology


This is just a bit of trivia, since the real issue here isn't the printer being used, but the fact the media is silver halide (chemical) photographic paper.  While the gamut of this process is less than current inkjet, these prints still offer stunning color, and using some technologies such as Kodak Metallic paper offer results that so far cannot be duplicated with inkjet.

We currently have almost 200 noritsu printers in operation using a very tight workflow and have no banding issues, so I can't offer any advice as to what a person might change in their workflow to predict and remedy this.  However, both labs have great service departments, and I'm sure they've run into this before.  A phone call may be just the ticket.

[Theodore]

While I believe they might have a Lamda or two,  Millers and WHCC use mostly a combination of Durst Theta and Noritsu printers. Millers facility is quite amazing, I believe I counted 20 of these printers in all lined up when I toured their main plant a few years ago.    The Durst Theta uses a LED/ fiber optics system as the imaging source instead of a laser which is what the lamda and most of the Noritsu's use.  I believe both labs also have a Chromira for extremely large prints, which also uses LED technology


This is just a bit of trivia, since the real issue here isn't the printer being used, but the fact the media is silver halide (chemical) photographic paper.  While the gamut of this process is less than current inkjet, these prints still offer stunning color, and using some technologies such as Kodak Metallic paper offer results that so far cannot be duplicated with inkjet.

We currently have almost 200 noritsu printers in operation using a very tight workflow and have no banding issues, so I can't offer any advice as to what a person might change in their workflow to predict and remedy this.  However, both labs have great service departments, and I'm sure they've run into this before.  A phone call may be just the ticket.

Thanks Wayne - interesting.  You're running 200 noritsu printers - are you associated with a commercial print lab?  

Regarding your suggestion, I think it's a good one.  I could see the banding faintly in the file and so when it turned up so accentuated in the print, I thought "goodness, my 8 bit sRGB file simply isn't printable with the post crop" but your right; I should give them a call and see if they have any suggestions aside from "don't use the post crop effect in LR".

[Wayne Fox]

Thanks Wayne - interesting.  You're running 200 noritsu printers - are you associated with a commercial print lab?

A chain of portrait studios called Kiddie Kandids ... each one has it's own printer, plus a central lab with several printers to produce large prints and other products.

I started a lab in 1980 (portrait and wedding),  and in 1987 bought this chain (4 locations in Utah) with my largest customer.  We operated the business until a few years ago when we sold it.  He retired, and  my current role is part time consultant ... (very part time).