but there's still matrox out there, and always the most linux friendly. Shouldn't they be getting more support?
Uh, ever tried to get any Matrox Pxxx card to work with Linux? There are drivers... if you run the right distribution and have the luck to actually get it to work. I found even the crappy binary only drivers from ATI easier to get to work, not to mention the nvidia drivers.
Most Linux friendly hasn't been true since the G500 (where is my dual screen support? how about TV out??)
Now let see, With the drivers from xorg all RV2xx based ATI cards work (thats upto the 9200 at least) and with the current development drivers the RV300 based cards also work (9600 etc)
Nvidia simply means using a proprietary driver, which isn't somethign I like at all, but I definitely have to grant them that their drivers work and that they work on about any linux distribution. What is more, they bpther to support slightly more excentric platforms, and the source component of their proprietary driver is complete enough to port it to similar but yet unsupported platforms (development versions of FreeBSD, NetBSD etc)
So.. who has the best support for Linux? Depends on how you look at it, but in none of the ways of looking at this I can see how the answer will be Matrox.
This is too bad, because in the time of the G400 and such, they did have the best Linux support indeed.
Hmm, I only have a '2.5G' phone (does GPRS) and it has a public IP (when connected). While most commercial content aimed at phones is overpriced, there is absolutely nothing that prevents me from making a http connection to my own webserver and serve things to my phone directly. I do pay for the gprs connection from my phone, but that is it.
Your problem seems to be a clueless provider that doesn't provide you with internet access, just their own little 'intranet'. It has however nothing to do whatsoever with 3G phones, if your provider (like mine does) would get you a public IP then it would be as usefull as the apps on your phone can make it (a browser on a tiny screen is not very usefull really unless you take it into account when creating content for it).
I use this often for reading slashdot on my 6 hour long monthly train trips between Utrecht and Berlin..
It could be implemented as an OS scheduler with 4 run queues that each get used in turn.
A somewhat similar concept (virtual cpus) is used in products like vmware tho for somewhat different reasons, and there the actual scheduling is done (for as far as I know) by the OS scheduler.
If you've been in the industry for any length of time, I hope you understand my reaction is rooted in the frustration caused by the erosion of credibility afforded to actual web developers, due to the aforementioned folks.
Definitely. Oh, and thanks for taking the time to reply, it is appreciated.
When you post to a mailinglist you do not know where it is going to end up unless you run the list yourself maybe (people can still forward it to anywhere they want). You way to opt out of this is to not send things to a mailinglist.
Many mailinglists are archived also, usually with a nice web interface. Those archives already get indexed by google, so the fact that it is going to end up in a gmail account possibly is a non issue.
I have a 9600se around, but in my workstation is actually a 9200. No idea why I thought I had the 9600se in there. At any rate.. I know I was using a rv280 and not a rv300 based card, and for that one the new drivers seem to do quite a bit in performance.
I think that's a genuine danger - that you up the requirements for Linux
Seeing how pretty basic 3D hardware can do this, and how it takes load off of the cpu, it is actually more likely to lower the hardware requirements for building a nice looking and responsive desktop with Linux.
I have played with those drivers (actually, I'm kinda stuck with the new xorg ati drivers and dri because the kernel I use doesn't work well with the 'old' dri module).
They are nice, tho fairly incomplete as far as r300 support goes. It is a real step forward tho, and the difference with pure software rendering is huge.
For those with slightly older cards, the difference is quite substantial as well. For the first time I have seen a 9200 perform at least good enough to be usable for things like enemy territory on linux (something that has never been a problem on anything from nvidia that is more recent then a tnt2 really, with their proprietary drivers that is, while the proprietary ATI drivers have always given me huge problems, preventing them from being usable alltogether for me). Depending on the application you can see upto some 40% increase in performance for 3D work.
I haven't used them enough with a 9600 to say anything about stability with that hardware, but stability with the 9200 I'm using right now (well, a 9600SE actually, but that is pretty much a 9200 for as far as drivers go) is excelent.
All in all, for 'low-end' 3D graphics on Linux, ATI has become a more serious contender with those drivers. The fact that they are open source makes them the prefered choice now unless I want something for more serious gaming on Linux (or FreeBSD usually in my case).
Don't blame the tool. It's not that the result of the save for web feature is inherently crap, it's that the user is selecting a compression rate that's far too severe, causing artifacting and all sorts of ugliness. The settings you're telling them to use will result in an image that's not suitable for web publishing...1280x1024 and low compression? That's gonna give you a huge (for the web) file size, for starters. And most users wouldn't be able to see the whole image, or worse, their browser may resize it, most likely to an odd fraction of the original size, resulting in that great jaggedy look. Unless you're doing the unthinkable and using element attributes to resize the image in the page. *shudder*
Heh, how about a webserver that can resize the image before sending it to the client? It really works well, and having the full scale pictures on the server means you can let the client chose what quality to use.
Photoshop's save for web tool could never be seriously accused of explicitly hiding what it does from a user. If anything, it's got to be overwhelming for most users because of the amount of control it offers. You can apply multiple compression rates to areas of an image using alpha channels, along with a dearth of other features, all right there in the right sidebar. At any rate, how is simply selecting "90% quality or better" disclosing the process to the user, or giving them any control?
Ok, I think I wasn't very clear there. Of course 'save for web' offers a zillion things that you can adjust, but there is nothing that makes it 'for the web', its just a usefull set of options and a previewer. As a matter of fact, all the things it has could easily be optional features in the normal 'save as' (just have an advanced button there)
The issue I am running into with photoshop users is that they think that just using 'save for web' will produce a file suitable for publishing on the web. Are those users at fault for not learning the tool they are using? sure, but as I pointed out a few lines back, there is no reason for it to exist to begin with.
So you are right that it is not hiding options so much as encouraging people to not use the tool properly.
Wow. You obviously don't understand how this whole compression thing works. You can't just assume that one setting will work for every image. So much of it depends on the content of the image, since what you're really deciding is "how much of this image can I throw away before it's noticeably degraded?" Of course, if you're just using the lowest compression/highest quality setting (bad idea for the web), this is moot.
If that is what you send to the browser, then its a bad idea, yes. It never occured to you that you do not actually have to do that, but rather, can decide on scaling and compression when a browser asks for the picture, and at that moment you can make sure the browser isn't going to scale it down because it didn't fit and that kind of nonsense. This is not suitable for all pictures, and depending on use can become a bit expensive on the server, but it is very usefull when used properly.
It also never occured to you that the web designer may want to scale the picture him/herself to have more flexibility while implementing the design?
You do want a high quality source for those things however.
I guess it's a good thing you're not a graphics artist, but given you're comments, I also have doubts about your web development expertise.
That is only because the possibilities for a different approach never occured to you.
First, it's not a prediction, but an analysis of current manufacturer's products. I won't argue that you couldn't do it another way, but since they don't, your argument has little bearing on the cameras people are actually using.
I think you were saying that they cannot and will not implement this, I took the later for a prediction.
Second, you're not downsampling from 10-bit to 8-bit in-camera unless you're demosaicing and converting to jpeg, in which case you're converting from the sensor's RGBG representation to RGB-triplets. All of the colorspace calculations and tonal mapping is occuring in high-bit math to improve accuracy before the final result is downsampled. It's not a simple lop-off-a-few-bits and here's your value process.
I understand that, and depending on the situation, this may be a good idea, do all processing in 10bit, and only downsample for the final picture if needed (and don't if you don't have to). As an expert on the subject, you no doubt realize the advantages of doing the math first, then downsampling, as opposed to downsampling your sensor data first, and then doing all of your bayer calculations at your final bit count.
Definitely, and if you can, keep it at 10 bits all the way, and let whatever needs to display this decide on reducing it to the colorspace and accuracy of the display device (being it a computer display, printed paper, TV or whatever)
At any rate, I guess I had a point somewhere..
- When representing a bigger dynamic range in the same number of bits, you lose accuracy - Try to avoid downsampling because it inherently introduces noice and reduces accuracy. The loss of this may be compensated by better accuracy during calculations, but there are other ways to get some of that (ie, use floats for any intermediate results)
With regards to analog implementation of 'headroom', you can prevent overloading the sensor optically. Once it got overloaded however there is nothing you can do about it digitally. Whenever you try to register analog data digitally, you get a much better result if you can manage to match the analog data to your sensor with regards to dynamic range instead of normalizing it digitally later on. It gives you a bit less control over what happens, but it also allows for getting a more accurate source to work with since it can use the full range of the digital representation without any up/downsampling, and it allows for a sensor to behave more like analog film with regards to 'gracefully failing on overload'.
The really, really nice thing about "Save for web..." is that, for the web designer, you can see exactly what the compression does before you save it, and it has superb tweaking options for the end result to be the perfect compromise between size and looks. The result need not be crap, you just have to know what you're doing.
Definitely, I really like the 'preview' it gives you there. I just wonder why that is not an optional part in the normal 'save as' dialog.
If a graphic designer uses "save for web" in the way I think you mean ("gee, i'm uploading this to a web site, so I should use save for web, right?"), they should really read more Photoshop documentation.
Agreed, and that is indeed what I am running into quite a bit.
Somehow I assume you know this already, and that I'm just preaching to the congregation here... oh well.
On a side note, I try out the Gimp once in a while, but I can never seem to get around the interface. I know that's because I've been using PS since version 3, and the workflow of Gimp is just different.
Yes, and at times it is just not very logical and cumbersome.
Still, some of the ways of doing things in the Gimp seems cumbersome and unnecessary. I'm talking about selections, copy/paste etc. And can you actually use polygonal selection in the Gimp? If it's there, I haven't found it. And that's the tool I use the most in PS.
Sure, magnetic lasso and freehand selection is fine, but there's nothing like the accuracy of zooming in to 600% or more to cut something out pixel by pixel by pixel:)
Copy/paste I find pretty easy (and am pretty fond of the 'paste as new' feature), its selection that can be a real pain. I haven't found the feature you are looking for but then, I'm not in much need of it so I didn't look too hard. I often need inverted selection by color, and both programs do that pretty well. At any rate, in my experience accurate selection in the Gimp is often between a bit bothersome and utterly painfull if you need some random shape in a busy picture.
I currently have the save for web dialog in front of me, and I fail to see an option there that isn't in the Gimp, tho at differnet places often. Yeah, you'll have to do it in a slightly different way, very inconvenient if you ae used to doing things the Photoshop way already.
I still do not see the point of this dialog however, there is nothing inherently different about graphics used on the web when compared to other uses. Yeah, all uses have their specific requirements for quality (resolution, color space and depth etc), so why not have a 'save for press' option as well? and how about a 'save for later printing on a inkjet printer' and don't forget a variation for a laser printer, save for screen (with options for different screensizes of course)
The one good thing I can say about the 'save for web' dialog is that it gives somewhat of a preview so you can make a guess at how it will look with the settings you are using. That said, seeing how browsers tend to scale pictures to whatever size fits them nowadays, and use pretty bad algorithms for that (nearest neighbor), the final result is often not going to be what you have on your screen there anyway (and you don't need a seperate 'save for web' dialog for that really)
When you 'need' that dialog, you seem to have learned what to click on to achieve a specific effect, without understanding what you are actually doing, and that is exactly the thing I have run into with quite a few photoshop users. They learned from tutorials, but have no real idea about what they are doing.
Oh, and please, a dialog collecting a bunch of options is not a 'module', or are people suggesting that untill this 'module' got introduced, Photohop was lacking in the area of color reduction and/or scaling, or didn't support jpeg or gif or such?
You're right. We should be more like Denmark. See? They're only hated around the world because they allow freedom of the press. Hell, those embassies needed to be rebuilt anyway.
And that for a country that at least to American standards has implemented a socialist system...
I only pointed out that you're painting Americans with a fairly broad brush
You see, I comment on the political parties in the USA, and somehow you see that as being negative about Americans? To me that sounds like you are looking for the enemy.
Let me tell you something, I can see clear differences between the following things:
1. The American people (too diverse to apply any kind of generalisation) 2. The USA as a country 3. The current government of the USA 4. Political parties in the USA.
Can you?
Just to be clear, I do not believe the political parties in the USA represent the people who live in hte USA very well, at best they represent some fringe minorities that happen to have the means to obtain and keep to power. The only thing I can blame American people for is still tolerating such a system, but then, I can blame many people around the world for tolerating the systems they live under, so there is no 'us vs them' in there at all.
Just wondering.. who is the 'us' that you are suggesting I belong to, and who are the 'them'?
Oh, you are thinking about me being part of the supposedly 'anti american' old Europe, and 'them' obviously being all Americans?
There is something that you should understand. Your enemies won't point out your weaknesses other then by using them to destroy you. Your friends will point them out so you can do something about it before it is too late.
Hmm.. I believe you replied to the wrong post...:)
Re:It's not just commercial interests with money..
on
NASA Science Under Attack
·
· Score: 0, Offtopic
I certainly don't want my tax dollars to support every candidate that fills in the right forms. I'd rather put my money behind campaigns that actually represent what I think.
Well, I can understand that point of view, but I'd like to give you some things to think about..
In my opinion, it is more important to maintain a democratic system then to maintain a specific candidate or party. Because of that, I want my tax money to first of all be used to keep the democratic system working, even if that means part of the money will fund campeigns that I do not support.
Also, when parties get their campeign money from their supporters directly, that just results in concentrating power there where the money is, ie, people or corporations wiht a lot of money can directly buy influence. That is not how a one man one vote system is supposed to work, there the money is supposed to be of no relevance at all.
I don't have the perfect solution, but to me tax money being used to fund a democratic system and the parties in it seems a better guarantee for a democratic system then effectively letting people buy power and cover it with a thin veil of elections.
Are you an American? If so, he's your President too. If not, why are you commenting?
I'm not, and why am I commenting? because I have an opinion about it. Being able to see what is happening there doesn't take havign a specific nationality really.
As far as knowing a lot of Christians who want to push religion on others...I know of quite a few Atheists and Agnostics who do the same. Atheists in my opinion have tried to take away the idea of a Merry Christmas from me for years.
Did you even try to understand what I wrote, or for that matter, read it?
I said I know A FEW who try to push their religion onto others, and many who don't.
I never ever suggested there are many Christians who try to push their religion onto others, but I did say there are some, and also that mr. G. W. Bush is one of them.
Don't treat someone as the enemy just because you disagree with them.
So, don't act as you do, act as you say eh?
However, I guess everytime I Jehova's witness knocks on my door I should start screaming that they are trying to push their religion on me. (your words, not mine)
Why scream about it? Usually they have no problem with you politely telling them you are not interested.
With all respect, it is you who is getting all upset and responding like anyone who says something you don't like being the enemy, so much that it prevents you from reading, let alone understanding what someone else is saying.
Why don't they actually stop using the word ideology and just say they hate him because he is a Christian?
I know a few christians who want to push their religion onto others with any means at their disposal. I know a lot of christians who think that is a really stupid idea.
Your president belongs to the first group, not the second. It would not be fair to al the nice and decent christians out there to hate them because a not too smart president keeps calling himself a christian also.
Being able to 'see' it if one wants to is nice, and I'd have been surprised if you couldn't set the compression quality at all. Hiding it behind 'profiles' is the issue here. Many a graphics artist I dealt with seems to believe that this 'save for web' stuff does things that they cannot do themselves, or wouldn't be able to learn or understand themselves.
It is creating 'magic' where there is none. If Adobe wants those profiles (they are definitely usefull, why think up defaults for everything yourself all the time), they would do good to call them something like 'defaults for web', and explicitly show what such a profile does.
rofl and you got modded insightful too,/. is so sad.
Heh, +5 insightfull is a bit much for that post indeed.
But it does look to me like I have a point that you are just missing or are unwilling to accept.
Without judging what you said (its your beliefs, and you are definitely entitled to those), I'd suggest you look a bit more at the things both parties do, and listen a little bit less to what people say.
Then, to an outside observer, calling any mainstream party in the USA 'left' is just too fucking hilarious. Compared to any real socialist party in any other part of the world, both democrats and republicans are pretty much right-wing to extreme right. I'm not even talking about supposed socialist parties with marxist or maoist affinities, those are not proper socialist parties to begin with, I am talking about the typical social democrat party as found in many western countries.
For an outside observer, the libertarian and green parties in the USA a least look a little bit different still, but never got far enough to let any actions speak for them. Differences between democrats and republicans seem pretty superficial and created only for the purpose of having some difference at all, at least when looking at actions and not at party rhetoric.
At times it makes me wonder how peopel can make an informed choice if all there is is mud slinging, meaningless thetoric, and typical us vs them psychology.
You will notice how much faster Konqueror is on a low-end machine. I'm currently playing around with a so called 'thin client', which mostly suffers from having less memory (256mb), a not too fast cpu (800mhz via p3 compatible) and little storgae (128mb). The cpu and memry matter for a browser.
Now, I can run either Firefox or Konqueror on this machine, in both cases without needing swap. Both will slowly grind to a hold after a couple of hours of intensive use due to running out of memory. No idea if it is Konqueror or some other kde component that it uses, but it seems there is little difference between the 2 when it comes to slowly wanting to eat more and more memory.
Konqueror renders a lot faster, and stays more responsive, untill the moment a page tries to use some kind of javascript. When pages do, Firefox just goes on, being slightly sluggish just like it already was, while at times Konqueror grinds to a virtual halt for a few secs, and is slow as hell untill another page gets loaded. Doesn't always happen, but often enough to prevent using it.
And yeah, I got both of them installed, but not on the 128mb internal storgae, both come from an external usb conencted disk.
Not exactly, they have patents on ways to do colorspace conversion however.
(Somebody's going to have to explain to my copy of Corel PhotoPAINT, which happily does CMYK, that it's violating an Adobe patent.)
Maybe, just maybe Corel licenses those patents from Adobe? Maybe they have found a way around it and can afford the risk of being sued? I don't know, but I do know that in the current situation, The Gimp does not have either option.
So a sensor can not, does not, and will not implement a tone curve in a "analog" way. When done in-camera it's a digital process, on digital data.
What it will or won't depends on market and manufacturors, I wouldn't make such a prediction.
It definitely can be implemented the analog way, and for anyone knowing a bit about converting and resampling digital representations of analog data it will be plainly obvious why it can and should be implemented there, if implemented at all. Doing this digitally is going to be both relatively expensive, and it will not come anywhere near the quality that you'll get from spending the same on an analog solution.
That an analog setup is not compatible with what some manufacturors are doing currently could be, but that in no way means it can't be done or that it won't make sense to do, and a lot with the fact that 'being all digital' is a marketing argument.
The extra data is not thrown away as just "noise".
No, and if you read a bit better, that was nto what I said.
A conversion from 10 to 8 bits can happen in a few ways:
1. You throw away the 2 least significant bits 2. You resample the data at 8 bits. 3. You divide every value by 4 (shift right twice)
You do one of those, not all at the same time.
In all cases, you reduce a range represented in 1024 'steps' to one represented in 256 'steps'.
Option 1. obviously throws away data. Option 2. introduces noise as an inherent consequence of the resampling process, no matter how good your asic or algorithm. Option 3. makes that you could as well just have used 8 bit data, and effectively is the same as option 1.
The only option that actually gets you some result of the extra data is option 2, and as said, resampling introduces at least some noise. I suggest to learn a lot more about digital representation of analog data, an explanation of why this is will be a bit lengthy for this post.
As long as the camera can produce the raw scanner data without doing any conversion or downsampling, it would be fine, you can use 10 or 12bit color during processing, and only downsample to 8 bit if needed for the final result.
Slashdot Mirror
Here, we get rock-bottom service at deluxe prices, including the aforementioned limited access.
Yes, and the providers are to blame for that, not 3G mobile phones.
but there's still matrox out there, and always the most linux friendly. Shouldn't they be getting more support?
Uh, ever tried to get any Matrox Pxxx card to work with Linux? There are drivers... if you run the right distribution and have the luck to actually get it to work. I found even the crappy binary only drivers from ATI easier to get to work, not to mention the nvidia drivers.
Most Linux friendly hasn't been true since the G500 (where is my dual screen support? how about TV out??)
Now let see, With the drivers from xorg all RV2xx based ATI cards work (thats upto the 9200 at least) and with the current development drivers the RV300 based cards also work (9600 etc)
Nvidia simply means using a proprietary driver, which isn't somethign I like at all, but I definitely have to grant them that their drivers work and that they work on about any linux distribution. What is more, they bpther to support slightly more excentric platforms, and the source component of their proprietary driver is complete enough to port it to similar but yet unsupported platforms (development versions of FreeBSD, NetBSD etc)
So.. who has the best support for Linux? Depends on how you look at it, but in none of the ways of looking at this I can see how the answer will be Matrox.
This is too bad, because in the time of the G400 and such, they did have the best Linux support indeed.
Hmm, I only have a '2.5G' phone (does GPRS) and it has a public IP (when connected). While most commercial content aimed at phones is overpriced, there is absolutely nothing that prevents me from making a http connection to my own webserver and serve things to my phone directly. I do pay for the gprs connection from my phone, but that is it.
Your problem seems to be a clueless provider that doesn't provide you with internet access, just their own little 'intranet'. It has however nothing to do whatsoever with 3G phones, if your provider (like mine does) would get you a public IP then it would be as usefull as the apps on your phone can make it (a browser on a tiny screen is not very usefull really unless you take it into account when creating content for it).
I use this often for reading slashdot on my 6 hour long monthly train trips between Utrecht and Berlin..
It could be implemented as an OS scheduler with 4 run queues that each get used in turn.
A somewhat similar concept (virtual cpus) is used in products like vmware tho for somewhat different reasons, and there the actual scheduling is done (for as far as I know) by the OS scheduler.
If you've been in the industry for any length of time, I hope you understand my reaction is rooted in the frustration caused by the erosion of credibility afforded to actual web developers, due to the aforementioned folks.
Definitely. Oh, and thanks for taking the time to reply, it is appreciated.
When you post to a mailinglist you do not know where it is going to end up unless you run the list yourself maybe (people can still forward it to anywhere they want). You way to opt out of this is to not send things to a mailinglist.
Many mailinglists are archived also, usually with a nice web interface. Those archives already get indexed by google, so the fact that it is going to end up in a gmail account possibly is a non issue.
Whoops, you are completely right of course.
I have a 9600se around, but in my workstation is actually a 9200. No idea why I thought I had the 9600se in there. At any rate.. I know I was using a rv280 and not a rv300 based card, and for that one the new drivers seem to do quite a bit in performance.
I think that's a genuine danger - that you up the requirements for Linux
Seeing how pretty basic 3D hardware can do this, and how it takes load off of the cpu, it is actually more likely to lower the hardware requirements for building a nice looking and responsive desktop with Linux.
I have played with those drivers (actually, I'm kinda stuck with the new xorg ati drivers and dri because the kernel I use doesn't work well with the 'old' dri module).
They are nice, tho fairly incomplete as far as r300 support goes. It is a real step forward tho, and the difference with pure software rendering is huge.
For those with slightly older cards, the difference is quite substantial as well. For the first time I have seen a 9200 perform at least good enough to be usable for things like enemy territory on linux (something that has never been a problem on anything from nvidia that is more recent then a tnt2 really, with their proprietary drivers that is, while the proprietary ATI drivers have always given me huge problems, preventing them from being usable alltogether for me). Depending on the application you can see upto some 40% increase in performance for 3D work.
I haven't used them enough with a 9600 to say anything about stability with that hardware, but stability with the 9200 I'm using right now (well, a 9600SE actually, but that is pretty much a 9200 for as far as drivers go) is excelent.
All in all, for 'low-end' 3D graphics on Linux, ATI has become a more serious contender with those drivers. The fact that they are open source makes them the prefered choice now unless I want something for more serious gaming on Linux (or FreeBSD usually in my case).
Don't blame the tool. It's not that the result of the save for web feature is inherently crap, it's that the user is selecting a compression rate that's far too severe, causing artifacting and all sorts of ugliness. The settings you're telling them to use will result in an image that's not suitable for web publishing...1280x1024 and low compression? That's gonna give you a huge (for the web) file size, for starters. And most users wouldn't be able to see the whole image, or worse, their browser may resize it, most likely to an odd fraction of the original size, resulting in that great jaggedy look. Unless you're doing the unthinkable and using element attributes to resize the image in the page. *shudder*
Heh, how about a webserver that can resize the image before sending it to the client? It really works well, and having the full scale pictures on the server means you can let the client chose what quality to use.
Photoshop's save for web tool could never be seriously accused of explicitly hiding what it does from a user. If anything, it's got to be overwhelming for most users because of the amount of control it offers. You can apply multiple compression rates to areas of an image using alpha channels, along with a dearth of other features, all right there in the right sidebar. At any rate, how is simply selecting "90% quality or better" disclosing the process to the user, or giving them any control?
Ok, I think I wasn't very clear there. Of course 'save for web' offers a zillion things that you can adjust, but there is nothing that makes it 'for the web', its just a usefull set of options and a previewer. As a matter of fact, all the things it has could easily be optional features in the normal 'save as' (just have an advanced button there)
The issue I am running into with photoshop users is that they think that just using 'save for web' will produce a file suitable for publishing on the web. Are those users at fault for not learning the tool they are using? sure, but as I pointed out a few lines back, there is no reason for it to exist to begin with.
So you are right that it is not hiding options so much as encouraging people to not use the tool properly.
Wow. You obviously don't understand how this whole compression thing works. You can't just assume that one setting will work for every image. So much of it depends on the content of the image, since what you're really deciding is "how much of this image can I throw away before it's noticeably degraded?" Of course, if you're just using the lowest compression/highest quality setting (bad idea for the web), this is moot.
If that is what you send to the browser, then its a bad idea, yes. It never occured to you that you do not actually have to do that, but rather, can decide on scaling and compression when a browser asks for the picture, and at that moment you can make sure the browser isn't going to scale it down because it didn't fit and that kind of nonsense. This is not suitable for all pictures, and depending on use can become a bit expensive on the server, but it is very usefull when used properly.
It also never occured to you that the web designer may want to scale the picture him/herself to have more flexibility while implementing the design?
You do want a high quality source for those things however.
I guess it's a good thing you're not a graphics artist, but given you're comments, I also have doubts about your web development expertise.
That is only because the possibilities for a different approach never occured to you.
First, it's not a prediction, but an analysis of current manufacturer's products. I won't argue that you couldn't do it another way, but since they don't, your argument has little bearing on the cameras people are actually using.
I think you were saying that they cannot and will not implement this, I took the later for a prediction.
Second, you're not downsampling from 10-bit to 8-bit in-camera unless you're demosaicing and converting to jpeg, in which case you're converting from the sensor's RGBG representation to RGB-triplets. All of the colorspace calculations and tonal mapping is occuring in high-bit math to improve accuracy before the final result is downsampled. It's not a simple lop-off-a-few-bits and here's your value process.
I understand that, and depending on the situation, this may be a good idea, do all processing in 10bit, and only downsample for the final picture if needed (and don't if you don't have to).
As an expert on the subject, you no doubt realize the advantages of doing the math first, then downsampling, as opposed to downsampling your sensor data first, and then doing all of your bayer calculations at your final bit count.
Definitely, and if you can, keep it at 10 bits all the way, and let whatever needs to display this decide on reducing it to the colorspace and accuracy of the display device (being it a computer display, printed paper, TV or whatever)
At any rate, I guess I had a point somewhere..
- When representing a bigger dynamic range in the same number of bits, you lose accuracy
- Try to avoid downsampling because it inherently introduces noice and reduces accuracy. The loss of this may be compensated by better accuracy during calculations, but there are other ways to get some of that (ie, use floats for any intermediate results)
With regards to analog implementation of 'headroom', you can prevent overloading the sensor optically. Once it got overloaded however there is nothing you can do about it digitally. Whenever you try to register analog data digitally, you get a much better result if you can manage to match the analog data to your sensor with regards to dynamic range instead of normalizing it digitally later on. It gives you a bit less control over what happens, but it also allows for getting a more accurate source to work with since it can use the full range of the digital representation without any up/downsampling, and it allows for a sensor to behave more like analog film with regards to 'gracefully failing on overload'.
The really, really nice thing about "Save for web..." is that, for the web designer, you can see exactly what the compression does before you save it, and it has superb tweaking options for the end result to be the perfect compromise between size and looks. The result need not be crap, you just have to know what you're doing.
:)
Definitely, I really like the 'preview' it gives you there. I just wonder why that is not an optional part in the normal 'save as' dialog.
If a graphic designer uses "save for web" in the way I think you mean ("gee, i'm uploading this to a web site, so I should use save for web, right?"), they should really read more Photoshop documentation.
Agreed, and that is indeed what I am running into quite a bit.
Somehow I assume you know this already, and that I'm just preaching to the congregation here... oh well.
On a side note, I try out the Gimp once in a while, but I can never seem to get around the interface. I know that's because I've been using PS since version 3, and the workflow of Gimp is just different.
Yes, and at times it is just not very logical and cumbersome.
Still, some of the ways of doing things in the Gimp seems cumbersome and unnecessary. I'm talking about selections, copy/paste etc. And can you actually use polygonal selection in the Gimp? If it's there, I haven't found it. And that's the tool I use the most in PS.
Sure, magnetic lasso and freehand selection is fine, but there's nothing like the accuracy of zooming in to 600% or more to cut something out pixel by pixel by pixel
Copy/paste I find pretty easy (and am pretty fond of the 'paste as new' feature), its selection that can be a real pain. I haven't found the feature you are looking for but then, I'm not in much need of it so I didn't look too hard. I often need inverted selection by color, and both programs do that pretty well.
At any rate, in my experience accurate selection in the Gimp is often between a bit bothersome and utterly painfull if you need some random shape in a busy picture.
Heh, that calls me fanboy...
I currently have the save for web dialog in front of me, and I fail to see an option there that isn't in the Gimp, tho at differnet places often. Yeah, you'll have to do it in a slightly different way, very inconvenient if you ae used to doing things the Photoshop way already.
I still do not see the point of this dialog however, there is nothing inherently different about graphics used on the web when compared to other uses. Yeah, all uses have their specific requirements for quality (resolution, color space and depth etc), so why not have a 'save for press' option as well? and how about a 'save for later printing on a inkjet printer' and don't forget a variation for a laser printer, save for screen (with options for different screensizes of course)
The one good thing I can say about the 'save for web' dialog is that it gives somewhat of a preview so you can make a guess at how it will look with the settings you are using. That said, seeing how browsers tend to scale pictures to whatever size fits them nowadays, and use pretty bad algorithms for that (nearest neighbor), the final result is often not going to be what you have on your screen there anyway (and you don't need a seperate 'save for web' dialog for that really)
When you 'need' that dialog, you seem to have learned what to click on to achieve a specific effect, without understanding what you are actually doing, and that is exactly the thing I have run into with quite a few photoshop users. They learned from tutorials, but have no real idea about what they are doing.
Oh, and please, a dialog collecting a bunch of options is not a 'module', or are people suggesting that untill this 'module' got introduced, Photohop was lacking in the area of color reduction and/or scaling, or didn't support jpeg or gif or such?
You're right. We should be more like Denmark. See? They're only hated around the world because they allow freedom of the press. Hell, those embassies needed to be rebuilt anyway.
And that for a country that at least to American standards has implemented a socialist system...
I only pointed out that you're painting Americans with a fairly broad brush
You see, I comment on the political parties in the USA, and somehow you see that as being negative about Americans? To me that sounds like you are looking for the enemy.
Let me tell you something, I can see clear differences between the following things:
1. The American people (too diverse to apply any kind of generalisation)
2. The USA as a country
3. The current government of the USA
4. Political parties in the USA.
Can you?
Just to be clear, I do not believe the political parties in the USA represent the people who live in hte USA very well, at best they represent some fringe minorities that happen to have the means to obtain and keep to power. The only thing I can blame American people for is still tolerating such a system, but then, I can blame many people around the world for tolerating the systems they live under, so there is no 'us vs them' in there at all.
Just wondering.. who is the 'us' that you are suggesting I belong to, and who are the 'them'?
Oh, you are thinking about me being part of the supposedly 'anti american' old Europe, and 'them' obviously being all Americans?
There is something that you should understand. Your enemies won't point out your weaknesses other then by using them to destroy you. Your friends will point them out so you can do something about it before it is too late.
Hmm.. I believe you replied to the wrong post... :)
I certainly don't want my tax dollars to support every candidate that fills in the right forms. I'd rather put my money behind campaigns that actually represent what I think.
Well, I can understand that point of view, but I'd like to give you some things to think about..
In my opinion, it is more important to maintain a democratic system then to maintain a specific candidate or party. Because of that, I want my tax money to first of all be used to keep the democratic system working, even if that means part of the money will fund campeigns that I do not support.
Also, when parties get their campeign money from their supporters directly, that just results in concentrating power there where the money is, ie, people or corporations wiht a lot of money can directly buy influence. That is not how a one man one vote system is supposed to work, there the money is supposed to be of no relevance at all.
I don't have the perfect solution, but to me tax money being used to fund a democratic system and the parties in it seems a better guarantee for a democratic system then effectively letting people buy power and cover it with a thin veil of elections.
Are you an American? If so, he's your President too. If not, why are you commenting?
I'm not, and why am I commenting? because I have an opinion about it. Being able to see what is happening there doesn't take havign a specific nationality really.
As far as knowing a lot of Christians who want to push religion on others...I know of quite a few Atheists and Agnostics who do the same. Atheists in my opinion have tried to take away the idea of a Merry Christmas from me for years.
Did you even try to understand what I wrote, or for that matter, read it?
I said I know A FEW who try to push their religion onto others, and many who don't.
I never ever suggested there are many Christians who try to push their religion onto others, but I did say there are some, and also that mr. G. W. Bush is one of them.
Don't treat someone as the enemy just because you disagree with them.
So, don't act as you do, act as you say eh?
However, I guess everytime I Jehova's witness knocks on my door I should start screaming that they are trying to push their religion on me. (your words, not mine)
Why scream about it? Usually they have no problem with you politely telling them you are not interested.
With all respect, it is you who is getting all upset and responding like anyone who says something you don't like being the enemy, so much that it prevents you from reading, let alone understanding what someone else is saying.
Why don't they actually stop using the word ideology and just say they hate him because he is a Christian?
I know a few christians who want to push their religion onto others with any means at their disposal. I know a lot of christians who think that is a really stupid idea.
Your president belongs to the first group, not the second. It would not be fair to al the nice and decent christians out there to hate them because a not too smart president keeps calling himself a christian also.
Oh Noes! So Hidden! So confusing!
/. is so sad.
Being able to 'see' it if one wants to is nice, and I'd have been surprised if you couldn't set the compression quality at all. Hiding it behind 'profiles' is the issue here. Many a graphics artist I dealt with seems to believe that this 'save for web' stuff does things that they cannot do themselves, or wouldn't be able to learn or understand themselves.
It is creating 'magic' where there is none. If Adobe wants those profiles (they are definitely usefull, why think up defaults for everything yourself all the time), they would do good to call them something like 'defaults for web', and explicitly show what such a profile does.
rofl and you got modded insightful too,
Heh, +5 insightfull is a bit much for that post indeed.
But it does look to me like I have a point that you are just missing or are unwilling to accept.
Without judging what you said (its your beliefs, and you are definitely entitled to those), I'd suggest you look a bit more at the things both parties do, and listen a little bit less to what people say.
Then, to an outside observer, calling any mainstream party in the USA 'left' is just too fucking hilarious. Compared to any real socialist party in any other part of the world, both democrats and republicans are pretty much right-wing to extreme right. I'm not even talking about supposed socialist parties with marxist or maoist affinities, those are not proper socialist parties to begin with, I am talking about the typical social democrat party as found in many western countries.
For an outside observer, the libertarian and green parties in the USA a least look a little bit different still, but never got far enough to let any actions speak for them. Differences between democrats and republicans seem pretty superficial and created only for the purpose of having some difference at all, at least when looking at actions and not at party rhetoric.
At times it makes me wonder how peopel can make an informed choice if all there is is mud slinging, meaningless thetoric, and typical us vs them psychology.
You will notice how much faster Konqueror is on a low-end machine. I'm currently playing around with a so called 'thin client', which mostly suffers from having less memory (256mb), a not too fast cpu (800mhz via p3 compatible) and little storgae (128mb). The cpu and memry matter for a browser.
Now, I can run either Firefox or Konqueror on this machine, in both cases without needing swap. Both will slowly grind to a hold after a couple of hours of intensive use due to running out of memory. No idea if it is Konqueror or some other kde component that it uses, but it seems there is little difference between the 2 when it comes to slowly wanting to eat more and more memory.
Konqueror renders a lot faster, and stays more responsive, untill the moment a page tries to use some kind of javascript. When pages do, Firefox just goes on, being slightly sluggish just like it already was, while at times Konqueror grinds to a virtual halt for a few secs, and is slow as hell untill another page gets loaded. Doesn't always happen, but often enough to prevent using it.
And yeah, I got both of them installed, but not on the 128mb internal storgae, both come from an external usb conencted disk.
Whu? Adobe has patented CMYK? My eyes! My eyes!
Not exactly, they have patents on ways to do colorspace conversion however.
(Somebody's going to have to explain to my copy of Corel PhotoPAINT, which happily does CMYK, that it's violating an Adobe patent.)
Maybe, just maybe Corel licenses those patents from Adobe? Maybe they have found a way around it and can afford the risk of being sued? I don't know, but I do know that in the current situation, The Gimp does not have either option.
So a sensor can not, does not, and will not implement a tone curve in a "analog" way. When done in-camera it's a digital process, on digital data.
What it will or won't depends on market and manufacturors, I wouldn't make such a prediction.
It definitely can be implemented the analog way, and for anyone knowing a bit about converting and resampling digital representations of analog data it will be plainly obvious why it can and should be implemented there, if implemented at all. Doing this digitally is going to be both relatively expensive, and it will not come anywhere near the quality that you'll get from spending the same on an analog solution.
That an analog setup is not compatible with what some manufacturors are doing currently could be, but that in no way means it can't be done or that it won't make sense to do, and a lot with the fact that 'being all digital' is a marketing argument.
The extra data is not thrown away as just "noise".
No, and if you read a bit better, that was nto what I said.
A conversion from 10 to 8 bits can happen in a few ways:
1. You throw away the 2 least significant bits
2. You resample the data at 8 bits.
3. You divide every value by 4 (shift right twice)
You do one of those, not all at the same time.
In all cases, you reduce a range represented in 1024 'steps' to one represented in 256 'steps'.
Option 1. obviously throws away data.
Option 2. introduces noise as an inherent consequence of the resampling process, no matter how good your asic or algorithm.
Option 3. makes that you could as well just have used 8 bit data, and effectively is the same as option 1.
The only option that actually gets you some result of the extra data is option 2, and as said, resampling introduces at least some noise. I suggest to learn a lot more about digital representation of analog data, an explanation of why this is will be a bit lengthy for this post.
As long as the camera can produce the raw scanner data without doing any conversion or downsampling, it would be fine, you can use 10 or 12bit color during processing, and only downsample to 8 bit if needed for the final result.