Photoshop plugins that haven't been updated for CS

[Dan Wells]

I consider the collection of plugins that don't work with CS3 on Intel Macs, such as Nik Sharpener and the Canon print plugin for the iPF5000 to be a significant problem, and largely the fault of the plugin makers. I'm sure there are many more problem plugins that I'm not aware of right now, but those two (especially Nik Sharpener) have really gotten on my nerves. At least the print plugin is, by definition, the last thing I do to an image - it's a pain to save up a bunch of images ready to be printed, then relaunch Photoshop in Rosetta (where the plugin works) annd print them, but at least it works that way. Nik Sharpener, on the other hand, is used at multiple styages of workflow, so it involves either:

Always running Photoshop in Rosetta to accomodate Nik Sharpener
Constantly quitting and relaunching Photoshop in and out of Rosetta
Giving up on Nik Sharpener (I have a demo of PixelGenius Sharpener, which I like, and which IS native)

Come on, Nik (and Canon) - this program was in PUBLIC beta for nearly six months, and I'm sure that developers have had CS3 for longer than that. Intel Macs have been around for nearly a year and a half, and Apple has sold no non-Intel Macs for close to a year.  It's no surprise that CS3 on a new Mac doesn't like PPC only plugins! In many cases, PPC to Universal is just a recompile, a one-day job.  More infuriatingly still, Nik is still selling Nik Sharpener for the Mac at a significant price, even promoting it as new software, without ever mentioning that it doesn't work correctly on any Mac sold in the last nine months. Before posting on a public forum, I wrote to Nik Software, inquiring about whether they had an update. Their response was "we're working on it, but will not quote a ship date - would you like more information when we're ready?"   Mac Pros are selling very well right now, and most current sales are with a copy of the new Creative Suite. Third parties have a responsibility to update their software, or at least to slap warnings across the ads for the old versions - THIS DOES NOT WORK ON NEW MACS.


                                                       -Dan

[Schewe]

In many cases, PPC to Universal is just a recompile, a one-day job.
[a href=\"index.php?act=findpost&pid=116155\"][{POST_SNAPBACK}][/a]

Well, YOU'VE been drinking Steve Job's Kool-Aid, huh?

Yeah, just a "re-compile" and a COMPLETELY NEW FRIGGIN' COMPILER THAT BARELY WORKS!

See, most all developers for Macs used CodeWarrior...CodeWarroir was developed by FreeScale, who also made PPC G4 chips (a spin-off from Motorola). So, you figure CodeWarrior would get rev'ed for MacIntel? Silly boy...

So, now ya GOTTA use Xcode from Apple (hey, it's FREE)...and the SDK for CS3 required Xcode 2.4.1 which was only availble last Sept because earlier versions were too buggy to compile.

So, now, not only do you have to port your CodeWarrior projects to Xcode, you also have to learn how to use Xcode...and then you have to get it to compile for Universal Binary (UB) so that it will run natively under PPC and Intel. And if you develop cross-platform your code must compile for XP & Vista too.

I'm not trying to make excuses for Nik...I'm just explaining that you don't have really have an "accurate" handle on the situation. It takes time & effort...something Apple likes to gloss over to sell you all those fancy new Mac Pros.

...although I will admit I'm a bit surprised that Nik ain't got their stuff done yet, as for Canon, I wouldn't be holding my breath...

:~)

[Dan Wells]

I may have been mistaken, but I hadn't realized CodeWarrior had hung on until recently - Jeff Schewe certainly knows a great deal more about Mac development than I do (I'm a 23 year Mac user, but no development experience apart from some sophisticated FileMaker in the late 90s). I had previously thought that a lot of CodeWarrior stuff had gotten moved over in the early OS X days (wasn't that why Adobe took a long time with an OS X version of Photohop?)
       Anyway, I'm more surprised at Nik than Canon as well - when I talked to Nik, they were very courteous, but weren't offering any timeframe at all (but suggesting that it will be a completely new version of Nik Sharpener, and that it isn't coming terribly soon). I guess I'll switch over to PhotoKit Sharpener.

                                 -Dan

[Mark D Segal]

Dan, I have zero insight into the innards of the issues Jeff has raised here, but let me contribute two thoughts: (1) you can get around the Nik Sharpener problem by buying PK Sharpener Pro. It works fine with CS3 as long as you select the Background Layer before launching it. I agree with Michael Reichmann's assessment that it is the best sharpening plug-in period, and I've experimented with a number of them. Of course we all have our favorites and in principle we would like to be able to use them seemlessly from one version of PS to the next, but obviously life isn't going to be that simple.

(2) On the Canon matter, I agree with Jeff. Hell, they haven't even produced a satisfactory manual for that highly complicated printer, let alone resolved other issues, and the kind of "attitude" they seem to have viz a viz their customers and support, widely complained about all over the web, doesn't bode well. When the IPF 5000 was first announced I ordered one to get around the ink-switching business on my 4800, but I cancelled once I saw the issues cropping-up, and have been following the matter since, with some dismay. Much as Epson makes excellent printers and I appreciate (marvel at) the very high quality coming from my 4800, viable competition is a good thing.

[Schewe]

I may have been mistaken, but I hadn't realized CodeWarrior had hung on until recently
[a href=\"index.php?act=findpost&pid=116573\"][{POST_SNAPBACK}][/a]

CodeWarrior was still the top compiler for the Mac right up till the point Steve Jobs announced that Apple was moving to Intel chips and dropping FreeScale and IMB as suppliers. Which was June, 2005. Even after the announcement, CodeWarrior came out with a major upgrade (to version 10) that had already been planned and a couple of bug fixes since. But since Freescale owns CodeWarrior, they were doing that for their customers writing for PPC, not for Apple.

At one point, CodeWarrior WAS cross=platform, in that you could do a CodeWarrior Project and compile for both Windows and Mac...but I think Freescale sold the Win compiler to Sony (for their Play Stations).

Xcode, is now at 2.4.x (I don't remember what the last update number was) and the first version that Adobe certified to compile Photoshop code was 2.4.1 released Sept 2006.

If you compare CodeWarrior (at version 10) with Xcode (at version 2.4.x) you will see exactly why it took Adobe this long to be able to migrate over to Xcode. Xcode, less that one year ago couldn't even handle a project as large as Photoshop. It simply couldn't compile it.

What you may be thinking of is Cocoa which was the Apple OS X development environment-which is cool but doesn't work well at all when trying to develop cross=platform since it depends so much on the Apple API's and is not designed for really big projects. I would be very surprised if Nik was using  Cocoa...Lightroom is though (which caused no end of problems when doing the Windows version and may be the root problem for why the Mac version is so much faster than Windows).

You are just another victim of the Steve Jobs "Reality Distortion Field" where what Steve says sounds really, really good, until you are faced with the cold, harsh cruelty of facts.

This stuff will all get done, eventually...but it sure wasn't as easy or painless as Steve made it seem.

The upside is stuff run REALLY, REALLY FAST on the new MacIntel stuff, once you get it out of Rosetta.

[Dan Wells]

Thanks, Jeff, for a very clear explanation of what's really going on behind the scenes. It seems like it's a lot more complicated than it appears at first glance to move stuff over to Intel. Apple released Intel Macs nine months before there was any compiler that could handle something as big as Photoshop?!? Sony owns the Intel version of CodeWarrior (which exists)?!? Why doesn't Sony turn around and release CodeWarrior for MacIntel (or sell the rights to someone who's interested)?
       Since you're one of the folks behind PixelGenius, I guess Nik's loss in this case is your gain - I've been evaluating PhotoKit Sharpener this past week, and am just about ready to buy it. I have to say that (in addition to great output), one plus to PhotoKit is the team of certified Photoshop experts behind it... I have two (very minor) feature requests - one is an "inkjet 600 dpi" output setting. This would accomodate the Canon plugin, which sends data to the printer at 600 dpi (as far as I know, it's the only thing that does) - I've been using inkjet 480 dpi glossy fine with nice results, but I'd imagine a dedicated 600 dpi setting might even be a bit better. In a similar vein, is there a need for a "digital ultra high resolution" setting in capture sharpening - for 12 -17 mp cameras, as well as 20+ mp backs?  The second request is for PhotoKit to set the active layer back to the background when it's done (I've tried to do a minor last minute correction after running PhotoKit, and been momentarily surprised when it doesn't work, then remembered that I'm in the wrong layer).
       You're right about native stuff really flying on Intel Macs - I've been processing 100-600 MB files in CS3 (EOS 1Ds mk II files open up to 95 mb PSDs, then Genuine Fractals resizing and PhotoKit layers increase size from there), and I've been shocked at how fast it handles most operations.


                                                      Thanks,

                                                      -Dan

[Schewe]

I have two (very minor) feature requests - one is an "inkjet 600 dpi" output setting. This would accomodate the Canon plugin, which sends data to the printer at 600 dpi (as far as I know, it's the only thing that does) - I've been using inkjet 480 dpi glossy fine with nice results, but I'd imagine a dedicated 600 dpi setting might even be a bit better.[a href=\"index.php?act=findpost&pid=116726\"][{POST_SNAPBACK}][/a]

Yes, there is thought about doing that. But I hope you're only sending "native" resolution at 600ppi? If the size of your image would produce enough real, non-upsampled rez, then it's possible that sending that at 600ppi might be useful. But it really woudn't be useful to upsample to get to 600ppi and then sharpen.

In a similar vein, is there a need for a "digital ultra high resolution" setting in capture sharpening - for 12 -17 mp cameras, as well as 20+ mp backs?

Bruce Fraser and I struggled with that back when the 1Ds MII was released...we tried to see if that was a worthwhile effort-and decided that it wasn't. But that was before a lot of the mega MP backs were released. While Bruce is no longer with us, his work will continue on. At some point (I don't know when) we will be doing an upgrade to PhotoKit Sharpener 2.0. We have some interesting things in the works...and not only in our own products, but I can say no more...

:~)

Since you're one of the folks behind PixelGenius, I guess Nik's loss in this case is your gain

Ironic isn't it? Me standing up for Nik? Delicious irony, that...

But I'm more concerned that all developers don't get painted black because Jobs understated (and hyped like heck) the "easy transition" to MacIntel. Don't get me wrong...it was the right thing to do for Apple, but they left _ALL_ of their developers twisitng in the wind-and it ain't the first time.

Apple has a long track record of surprising, total changes in tech. 68K to PPC, OS 9 to OS X, and now PPC to Intel. All of these were like major sex change opperations for developers and yet, we all continue developing for the Mac.

The company that got hit worst by the Intel switch was Adobe...contrary to speculation, Adobe literally found out on the Fri afternoon before the Mon morning keynote. Jobs sent a limo for Bruce Chizen and told him. I gotta tell you there was total friggin' chaos that weekend.

In June 2005, Jobs stated that they would ship their FIRST MacIntel within 1 year (June 2006) and finish the transition by the _END_ of 2007. Well, that changed a bit (largely because IBM and Freescale basically quit making chips for Apple) and Apple moved up the ship date of the first MacIntel to Jan 2006 so they would have some computers to sell (kinda important if you're a computer company).

And, with the shipping last fall of the Mac Pro towers, Apple has completed the transition. Remarkable for Apple (and Intel) but the change in schedual really screwed with software developers cause Apple started shipping hardware before Universal Binary versions of software could be written. Literally, the first UB version of Photoshop CS3 was built in Sept, 2006. And it couldn't be built (compiled) until Xcode 2.4.1.

So, yeah, give the software devs a bit of a break...this stuff happened WAY faster than Apple said it would. But, it's water under the bridge-as soon as the devs get the UB version released...

[mistybreeze]

But, it's water under the bridge-as soon as the devs get the UB version released...
[a href=\"index.php?act=findpost&pid=116728\"][{POST_SNAPBACK}][/a]

Unfortunately, many of us have drowned in that water...waiting, as we tried to earn a living with a workflow that doesn't work and flow.

Thank you, Jeff, for being a level-headed diplomat. It's always nice to read opinion that's balanced and fair, a rare thing these days.

It's not natural for competing businesses to "talk" with one another and since Adobe does not make computers...

One might think that Adobe would get more respect from Apple. Apple loyalists have built many, many careers using Photoshop. Frankly,  Photoshop is a photo-edit application worthy of everyone's respect. No one company can do it all.

I tend to blame these things on men and their egos. Maybe I'm too sensitive for big business but in the computer/software world, it seems to me that customers benefit most when companies talk to one another. And when customers benefit, sales become a no-brainer. I'm blonde, so, what do I know.

[Dan Wells]

Great to get a Mac history lesson from one of the people who really knows the inside story. Adobe really did get left twisting in the wind by Apple. How far in advance did APPLE know? Did IBM and Freescale drop an untenable processor roadmap on them with little warning? I know there was huge frustration over G5 class portables and with the speed of the G5 stalling out (although, ironically, nobody seems to have shipped processors much over 3 gHz except for the old Netburst Pentium architecture, which was notoriously slow per clock cycle).The Core 2 seems like a nice architecture, though...
     On the question of interpolating to 600 dpi (maybe we should move this to the printing forum, or post a summary over there), I'm wondering what's best. I'm using a 1Ds mk II, so I have a lot of resolution, but not enough to print anything native at 600, apart from small prints. The Canon iPF 5000 plugin interpolates anything it's fed to 600 dpi (it has an option for 300 as well, but that's greyed out on many papers), and it does that archaically - it has a choice of nearest neighbor or bilinear. If I want to print 16x24, I have slightly over 200 native dpi. Is my best choice:

1. Don't interpolate, run PhotoKit Sharpener for 200 dpi, let the plugin rez it up in bilinear
2.Interpolate (Genuine Fractals) to 300 dpi, run PhotoKit Sharpener for 300 dpi, let the plugin rez it up in bilinear
3.What I have been doing - Interpolate to 300 in Genuine Fractals, run Photokit for 480, let the plugin go to 600 - This is a much smaller file than letting GF go to 600, and I haven't been getting oversharpening by running PhotoKit for the higher resolution (anticipating what the plugin will do)
4. Interpolate all the way to 600 in Genuine Fractals (or some other technique/combination), PhotoKit at 480, keep the plugin from interpolating at all. A 600 dpi (at 16x24) Genuine Fractals file really looks interpolated on screen, while the 300 dpi file looks very good. I thought that the 600 dpi file looked worse at 50% than the 300 dpi does at 100%, the one time I tried letting GF go all the way to 600 - I didn't print that file, though.


                                                          -Dan

[Mark D Segal]

Dan,

Jeff has much more scientific background in these applications than I ever will, but just because I don't know what I'm talking about won't prevent me from making a few observations:

(1) Bicubic sharper/smoother vs Genuine Fractals: Some say the the former replaces the need for the latter completely. True? Would that simplify your workflow?

(2) Using PK Output sharpener: I've engaged in discussion with them about matching file PPI with their settings, or using file sizes in-between without resizing to match. The general conclusions from that discussion are that the closer the match the better but I won't be the worse off for wear using the closest match they offer within their ranges relative to my file resolution - going higher for sharper results and lower for a bit less sharper results. We weren't discussing anything like a jump between 600 and 480 however.

(3) There is no substitute for doing your own tests MAKING PRINTS. Viewing sharpness on a monitor can be misleading as you probably know. It's only when you try various combinations and see the results coming out of the printer that you will know what set works best for you under your conditions. I've done it by setting up a matrix of all the combinations I'm interested in evaluating, and just printing them out one after another. Then I look at the prints without a loupe. The key is to select a test image that has low noise and a lot of texture and fine linear detail.

[thompsonkirk]

Mark, as you might guess, many of us have made such test prints already.

1.  For the only time I've ever been aware of it, Jeff is wrong:  The iPF5000, printing full-frame  '16 bit' 5D files on 17x22 paper (with careful use of PK capture & output sharpening), consistently produces sharper results & fewer artifacts if you res up from native res to 600 ppi.  That's how the plug-in was designed to work.  

2.  Resing up from native res to 600 ppi with either BiSmoother or another engine produces clearer results than letting the plug-in do the job itself.

3.  Resing up with BiSmoother does not do as good a job as Genuine Fractals - the results are less sharp & clear.

4.  GF is a slow program.  If you have limited memory & processor speed, you can get next-to-best results by using GF to res to 300 ppi, doing your local contrast enhancement & output sharpening, & then resing with BiSmoother (which is much faster) the rest of the way to 600 ppi just before printing.  Yes, this is observable without a loupe in the print size in question, though it's contrary to all the wise advice that you hafta sharpen last of all.

5.  If you have a fast processor & plenty of RAM, the best bet of all with a Mac & iPF5000 is to flatten the image, res up to 600 ppi with GF, do LCE, use PK output sharpener, and then flatten the image again.  (With the files & print size in question, it will have ballooned to 1.65GB).  Export the resulting '16-bit' file of about 550MB to the printer via the plug-in.  

My print tests indicate that's the best workflow - but then you run smack into the problem Dan & I & others have discussed on the iPF5000 Wiki - Canon's failure to provide a UB version of the printer plug-in for wide format printers & CS3.  You have 2 clumsy choices: (a) save the file, reopen CS3 in Rosetta, & print via the plug-in; or (2) close it & reopen in CS2 for plug-in printing.

We initially heard from Michael that printing 16-bit files via the plug-in produced superior results to using the printer driver.  This has subsequently been disputed - but many (most?) of us who listened have a stack of profiles that work with the plug-in & are not accurate with the driver.  So we do need an updated plug-in.  

Purchasers have written over & over about Canon's support failures, and as Jeff said, we're likely to asphyxiate by holding our breath for them.  They've apparently had all their software folks working on new models, without fixing what was clearly needed for the old ones.  But I would urge that nobody buy the new model until they take care of basic support for the previous one.  In the area of customer service they just aren't to be relied upon.

PS, Another little pitch for PKSharpener:  Don't try to use GF in combination with Nik.  You'll often generate artifacts that look like crinkled tinfoil in detailed areas.  

Kirk

[Schewe]

3.  Resing up with BiSmoother does not do as good a job as Genuine Fractals - the results are less sharp & clear.

[{POST_SNAPBACK}][/a]

Uh huh. . .and GF, as part of the uprez algorithm, has a pretty strong sharpening component (that's actually the part people seem to either like or hate)...

As to disputing the fact that uprezing beyond "native" resolution goes, that's not been my experience unless you are talking MAJOR uprez and post uprez manipulation as I outlined in this article [a href=\"http://www.digitalphotopro.com/articles/2005/septoct/artoftheupres.php]The Art Of The Up-res[/url] for DPP. But to really see any bennies, you need to do a bit 'O work after uprez.

I will be the first to tell ya I don't know squat about the Canon or HP printers either...I use Epson (cause they pay me to and I like them best). But it is my experience that Canons and HPs are "softer" printers than EPsons in the same class. In large measure because of the way that the respective drops are formed...

[Clive Carpenter]

Uh huh. . .and GF, as part of the uprez algorithm, has a pretty strong sharpening component (that's actually the part people seem to either like or hate)...

As a matter of interest, the latest version of GF (version 5.0) allows for sharpening to be either on or off and can be adjusted to taste when on.

It also allows for "film grain" to be added; looks like they read your article, Jeff  

Clive