britt_park

I've just recently got an Epson 3200. In my unscientific tests, scanning 6x4.5 Tri-X,

I'm getting a "real" resolution of slightly less than 2400 dpi. That is, scans at 2400

dpi, when USM'd, show a very close to sharp edge at the single pixel level. So in

practice I think you'll find the 3200 quite adequate for anything but really enormous

enlargements. My medium format scans are close to a full order of magnitude more

detailed than my 35mm scans on my Dual III.

Carl,

 

For most purposes including image processing I think there's not much to choose in

terms of stability between MS and OS X. I've been a software developer and have

used both OS X and linux for my own amusement and Windows 2000 and Solaris

because those were what the customer wanted. Under the kind of stress I put my

machines through, many compile, debug loops, installation and deinstallation of my

company's software, I had to to reinstall my Windows boxes about once every 6

months. Not something I've ever had to do with unixoid machines. I think the two

weak link in the various windows derivatives is the registry which seems still (as of a

year ago) to get corrupted easily and the large number of proprietary network

services that MS uses; crackers are still turning up new exploits frequently. But

software

development is not at all like digital image processing, and I would expect that your

average creative type will have no worse a time with XP as OS X or Linux (if they feel

adventurous). There's still the virus issue though, not much of a problem for those

who follow proper computer hygiene, but the unwary do get bitten a lot. For practical

purposes I agree with Carl. You can't go wrong with anything now (except cheapo

hardware).

There is only one really compelling reason to use a Mac and that is that OS X is unix

and unlike MS operating systems will not self-destruct over time. But even that

argument is getting less compelling since later versions of NT (Windows 2000, XP) are

really quite stable. You certainly get much more bang for your buck with PCs.

<p>Having never seen the Photoshop codebase, I can't be sure, but in principle it's

best to do everything you can in 16 bit mode, even if the original is only 8 bits. Each

manipulation produces round off error, which will be well beneath human

detectability in 16 bits but possibly notable using 8 bits (depending on how complex

your manipulations). The difficulty of course is that you have far fewer tools in 16 bit

mode. One helpful trick I learned is that if you save your 16 bit image and then

convert to 8 bits, and then re-open the original 16 bit image, you'll be able to make

magic wand selections with the 8-bit image and copy the selection to the 16 bit

version, for manipulations. I don't do much color work and have never run into color

shifts in 16 bits, but that's probably because I wasn't paying attention. My color

workflow is usually "levels" -> "color balance" -> Dodging and burning using the

lasso and magic wand selections on an 8 bit copy -> 8 bit (If I need to do something

that's only available in 8 bits). I remain puzzled by Adobe, for not making everything

work in 16 bit mode. Perhaps somebody else has heard their rationale.</p>

 

<p>- <a href="http://www.finephotography.org/brittpark/">Blatant Plug</a></p>

<p><a href="http://www.finephotography.org">Fine Photography</a></p>

If I remember correctly there are some conditions (of temperature perhaps?) in which elemental sulfur can be precipitated out of thiosulfate. Is the floating stuff really white or is it slightly yellow? If yellow I would suspect sulfur precipitation and would start over. I'm just blundering in the dark here. I'm cautious with photo chemistry and generally chuck out anything that's behaving strangely.

If you do use a Jobo in a motel, don't forget the old turn on the shower hot water for 5 minutes trick, to precipitate any dust. It worked well for me in my apartment dwelling days.

Wild guessing here. Is it possible that the holder was not seated completely for those two shots? Say if you had a grain of sand in the wrong place. The consistency of the light leak is what made me think of that. Otherwise I'd tend to suspect a defect in manufacture or processing for those sheets.

I'm sorry. I should never have started that other thread or this thread or followed up on either. If it's any help I feel pretty rotten.

I'm sorry. I should never have made the original post and should not have followed up. If it's any help I feel pretty rotten.

Good god, what a bunch of reactions to a simple question. Thank you Brian for answering. I do find it odd that my question has garnered anywhere from the suggestion that because I've only been a member for a short time I shouldn't ask questions to straight ad hominem attacks. What bearing does being new to photo.net have on what I should post? And what bearing does the quality or lack thereof of my photography have to do with the question in hand? The implied rules of any public forum would include: always being polite, not making verbal attacks, reading the archive before posting, all of which I diligently follow. It is not assumed in any other forum that I'm aware of that newer contributers are expected not to ask certain kinds of questions. If such subtle rules exist here on photo.net they should be prominently displayed.

Bad form following up one's own post but I noticed that somebody just asked a similar question to mine. I apologize for the overlap. I do think a prominent "manifesto" would be a good thing.

Is there a detailed statement of purpose for photo.net? The Terms of Use give no clues nor does About Us. If no such document exists, may I suggest that one be written. I would prefer a constitution that promised to run the sight as democratically as possible. But even one that stated that all decisions are made at the whim of those in charge would be useful. People would know what to expect.

I'd be interested in participating. When you work out the system to set up ones adds let me know.

<p><em>Unless something has changed, photo.net relies heavily on transactions and other SQL constructs that MySQL doesn't have (although they're working on it). Also, if you've ran Tux much or read the Tux mailing lists you'd know it's had little support recently and has some problems with lock-ups. It's also only good with static content and passes other requests off to Apache on another port.</em></p>

 

<p>Mike,</p>

 

<p>I never have run Tux and was unaware that it had issues. Does photo.net net really use lots of Oracle specific features, such as stored procedures, and does it really rely on strict transactional semantics? The DB traffic for photo.net must be 99% read access. To get "good enough" concurrency control for the 1% modify traffic I would think that simple "lock table"s would do. MySQL is wicked fast even with an occasional "lock table". If the photo.net people are of the "it has to be strictly transactional" school, MySQL as of 4.0.12 has strict transactions.</p>

 

<p>The more I learn about the way photo.net works the more I realize that it is woefully inertial, locked into very specific and unfortunate technologies. It might make sense to do something like set up a parallel slashcode implementation of photo.net as a movement into a sustainable future. Oracle, tcl, and an obscure web server are not a sound foundation for long term maintainability.</p>

 

<p><em>Support is key, too. I use mostly Dell servers with Red Hat Linux. I'm an enterprise customer with both. I also use some Sun equipment. Guess who has the best support when I have a problem? Sun. I have a Dell 2550 server with a random reboot problem and 6 months later I'm still trying to convince Dell there is something wrong with it. Sun just sends someone out to replace parts until the thing works. Dell and the other providers of enterprise-level Linux hardware need to work on this.</em></p>

 

<p>Your experience is sure different from mine. I have found Sun to be a most recalcitrant vendor. I've always wondered why they won the Unix Wars. Their hardware has always been behind the curve in performance and their service is dodgy. My service needs were unusual, I must admit, I worked on a product that lived partially in the Solaris kernel and so my questions were rarely in the on-line help available to their reps. Linux is pretty much self supported. If you stick with solid open source technologies, Linux just works. The hardware dies every once in a while (about 1 machine in 25 per year is our experience at my current lab with generic hardware) but at $500 a node and with most failed nodes still under warranty, that's less than any service contract would be.</p>

 

<p><em>BTW, instead of linking to Philip's Panda book you should link to the revision: Internet Application Workbook. The key concepts are the same but it uses examples of newer-tech things like C# and .NET.</em></p>

 

<p>Yoiks, C#, .NET. If you had mentioned XML I'd probably have had a heart attack. (Actually I don't have that much against XML, except for it's origin in SGML. The SGML folks should have been kept at least 100 miles away when XML was designed.) For web applications the only things that make any technological sense to me are the tried and true: Apache, cgi, perl, php, MySQL, and perhaps tomcat (Java is an aesthetically pleasing language but continues not to live up to its potential in terms of performance and technological stability. Sun introduces a new API for Java practically weekly without bothering to make solid, fast, free implementations of the core language available for all platforms.)</p>

 

<p><em>IMHO, ROI has little meaning for photo.net -- there's little investment to be had. In a perfect world I'd probably recommend photo.net go the Apache, MySQL, perl/python model, too, but not for technical reasons -- being a volunteer effort it'd be easier to find folks willing to work on the site. The sad reality is that people with experience in the technologies (i.e. Oracle) behind photo.net are few and in demand and busy working. The sheer amount of work to re-architect photo.net is vast -- I don't see how that could happen unless a rich donor or a grant shows up.</em></p>

 

<p>I agree with ROI being not that relevant to photo.net, though, whoever actually owns the site might disagree. I don't think re-architecting photo.net would be as bad as you make out, since there are several bulletin board systems freely available that could be adapted to photo.net's needs. Anyway, Philip Greenspun says that undergraduates at MIT complete web-based projects as complex as photo.net within a single course.</p>

 

<p>After my initial shock at the nasty responses I got from my initial post I've come to enjoy this thread immensely. It may not be photographically related, but it does provide some interesting and informed ideas for the photo.net elves.</p>

 

 

The basic trick is that you have to bow the film a bit to get it to feed into the reel slots. A gentle pinching pressure on the film as you wind will get you what you want. Unfortunately this is one of those things that is far easier demonstrated than described. Also, although it's not particularly likely, you may have a bent reel, which is preventing your success.

 

Keep practicing in daylight until it becomes second nature, then practice in the dark until it becomes second nature. It usually doesn't take all that long to get the hang of it.

<p>What is the ROI on photo.net's current setup? I don't know. I imagine not a great deal as only a small percentage of users are patrons, and banner add rates are not very high these days. If one could move photo.net to a new software and hardware system with little to no effort, the running costs amortized over several years would be less and thus the ROI would be greater. That is supposing a big if. Brian's post makes it pretty clear that photo.net is kind of backed into a software corner. Take a look at <a href="http://philip.greenspun.com/panda/index.html">this</a> to get an idea as to what kind of software corner photo.net is in.</p>

 

<p>Unfortunately the above sounds alarmist. However, for what photo.net does its current setup seems adequate if a little slow sometimes. As you rightly point out "new tech" becomes "old tech" very rapidly in the world of computers and software kind of the way that digital SLRs become obsolete so quickly. There is however a difference between good "old tech" and bad "old tech". Good "old tech" can be changed into "new tech" without much difficulty. Bad "old tech" can't. Thus in my fairly informed opinion photo.net is based on bad "old tech". I'm not worried because photo.net seems to be able to trundle along with what it has. But as a hedge against what the future may bring I'm going to donate some more money to photo.net.</p>

 

<p>- <a href="http://www.finephotography.org/brittpark/">Blatant Plug</a></p>

<p><a href="http://www.finephotography.org">Fine Photography</a></p>

<p>I think this is fun.</p>

 

<ul>

 

<li>a.) 7. Only a small percentage of my images work aesthetically in the end.</li>

 

<li>b.) 2. I'm not a story teller. I try to make photographs that please, emotionally, intellectually, humorously, or whatever.</li>

 

<li>c.) 5. I mostly shoot things that hold still. But when I do shoot something moving I get it about 50% of the time.</li>

 

<li>d.) 8. Negatives are a no brainer with a Jobo processor. I'm a solid competent, if inefficient, printer. I'm printing digitally now mostly.</li>

 

<li>e.) 8. I don't know if I have fire. But I still love all aspects of making pictures.</li>

</ul>

<p>I don't think this is too busy. You have a handful of shapes and textures in a pleasing arrangement. To me the bushes at the bottom are necessary and anchoring. Note that they are arranged as two clumps on the bottom corners, which gives a pleasing base for the composition. Those two clumps form a pleasing triangle with the arch within the arch.</p>

 

<p>Too many people, I think, don't give themselves enough credit. This is an excellent photograph and you should have no compunction proclaiming it so to the world.</p>

 

<p>- <a href="http://www.finephotography.org/brittpark/">Blatant Plug</a></p>

<p><a href="http://www.finephotography.org">Fine Photography</a></p>

My Dad.

<p>Mike,</p>

 

<p>I stand corrected about the origin and current status of AOLServer. It is not a piece of technology I would choose to use. I stand by my assessment of tcl. Face to face with perl it just doesn't stack up. I do like TK a lot. It's the only cross platform gui that really works well. If I wanted ultrafast static web-server performance I'd go with tux. Otherwise, I'd take the minor performance hit of Apache for its convenience, mod_perl, tomcat, mod_php, plugins, and its ready support.</p>

 

<p>I don't know the price of Oracle licences but I imagine they are obcenely expensive. If photo.net were to replace it's database server with commodity hardware I'd hope they would choose MySQL, which is wicked fast and much less expensive.</p>

 

<p>It's definitely not TCO that direct the IS practices of Fortune 500 companies. I'm not sure exactly what it is. TCO has been shown in many areas, in particular web-serving, to be lower with free unices. Yahoo serves with FreeBSD and runs the backend of their operation on a mixed bag of free unices. I think what does drive IT decisions, includes fear, (nobody gets fired for buying Sun (Microsoft)), CYA, (with a commercial product you always have someone to blame), and inertia, (often justifiably, with legacy systems in place that work who wants to change).</p>

 

<p>There is only one way that I know of that Big Iron unices make sense still. If you need to run a high bandwidth service (many thousands of transactions a second) and the service or software written to provide the service is not distributable across multiple servers, then an E10000 (or HP or IBM) with 64 processors and multiple fiber connections is the only viable choice. Banking, retail chains, financial market systems fit that bill. I do understand, shudder, however, that Nasdaq runs on NT. Their trading system was designed from the ground up with redundant commodity hardware in mind.</p>

 

<p>Obviously all of the above is my opinion. I worked for Interwoven, which bought Scriptics (Jon Ousterhout's company) and Interwoven got a number of fine programmers who plainly loved tcl. They were the first fine programmers I'd ever met who didn't subscribe to the aphorism "tcl, syntax as lovely as the bourne shell, with the expressive power to match". The ex-Scriptic programmers did some nifty internal projects with tcl. They did not succeed at getting a single non-Scriptics programmer to use tcl. I offer this as evidence only. For me tcl just doesn't work but Im all for variety and have nothing against people who love tcl. I love scheme so much that I've implemented it myself. But there aren't many people running to the banner of "lisp will save the world". We all have quirks.</p>

 

<p>I've tried to be non-personal here. If what I've said seems to anyone like I'm dissing them or their opinions please put it down to a lapse in my prose style.</p>

 

<p>- <a href="http://www.finephotography.org/brittpark/">Blatant Plug</a></p>

<p><a href="http://www.finephotography.org">Fine Photography</a></p>

<p>I put an inline particulate filter after the hot/cold mixer when I built my darkroom. I'm not sure it was really necessary but better safe than sorry makes sense to me. I use filtered water for everything but the final photo-flo rinse of negatives, for which I use distilled water.</p>

Brian et al., I did not intend to be mean spirited or "snide". Apparently I was. I therefore apologize for my original and follow up posts. My offer of help still stands, although it would take me a month to become conversant with tcl. I would be more help if you needed a new image server or other core services help. That's where most of my experience is.

<p>Gordon, you were actually disagreeing with Ivar and agreeing with me. A port from solaris to any free unix is trivial. I made an offer to Brian off-line to help with any programming difficulties. I agree with you that software maintenance costs much more than hardware. That is why I was advocating moving to a hardware platform that will be easier to maintain. If you doubt my qualifications read my <a href="http://www.sciencething.org/cv.pdf">CV</a>. </p>

 

<p>Could someone explain to me what I said that got so many people angry? I got royally peaved at one poster but had just enough restraint to flame off-line. I honestly don't think I said anything inflammatory, except for my tcl bashing. Anyone who feels like it can bash perl (my preferred scripting language) and I promise not to take offense.</p>

Actually in the long run switching to commodity hardware is likely to be considerably less expensive. Nodes are dirt cheap. Disks are dirt cheap. What does photo.net pay yearly for a maintenance contract on the E450 and RAID box? Also the problem with RAID boxes is that they max out and then you're left with a really expensive upgrade path, another RAID box, and expensive disks.

 

Please don't think I mean my comments harshly. I intended an implied grain of salt. I just thought you might be amused by some computer blather for a change of pace. I know how much work and money goes into keeping a high volume web-site going.

 

I worked at a startup that sells enterprise software and in the early days worked with a lot of customers, mostly Fortune 500 companies, and was never able to understand why most of those companies used expensive Sun hardware, when linux,freebsd,openbsd were so much cheaper and more reliable, not to mention easier to program for. To this day my company doesn't have a linux port because big companies just don't ask for it. I remain confounded.

<p>I just checked and www.photo.net seems to be running AOLserver a version of the Netscape web server. I also understand that your fronting the site with a Sun E450. These facts explain why www.photo.net is so tetchy and often slow in response. Why not ditch the E450, buy about 4 AMD Dual processor machines running FreeBSD to front the web serving with Apache, and rotating DNS. For your backend you'll need one whoopa machine (AMD x 4 + 400Gig running linux) for running the database and several dozen commodity PCs jammed each with 4 x 100G drives for image storage and serving. It'll be a much more robust and higher performance setup especially if you do data duplication.</p>

 

<p>Oh, and for Pete's sake, why do you use tcl? ;) a sillier scripting language I've never seen and I used to work for the company that bought Scriptics.</p>

 

<p>Cheers,</p>

<p><a href="http://www.sciencething.org/cv.pdf">Britt</a></p>