I am no developer but for years we used gimp to edit high resolution large imagery sets. Orthorectified aerial photography for GIS and engineering applications. We ultimately moved to Photoshop even against my wishes since I thought the majority of our issues could be solved if we persistently addressed them. Management on the other hand saw it in another light.

NOTE: I am not debating the merits of GIMP vs PS. Im pointing out real problems within the bounds of the question that has been asked.

Quoting Mark Shuttleworth :

- Is the GIMP a good platform to build on to try to achieve these goals? - What functionality would need to be added to the GIMP to challenge Photoshop?

First thing is what I think is memory management. GIMP can not deal with large rasters. Im talking 500MB and up. It even has issues with smaller ones. Anyone that disagrees with this obviously has not seriously tried to edit these types of files for hours. Yes, we have machines with the required ram, cpu, disk, swap, etc. Versus Photoshop... There is no comparison. PS 7.x works and works well. GIMP just doesnt. And yes I tried cinepaint and FilmGIMP. No go. And no I have not tried compiling gimp for x86 64bit but then whats the point? PS works on 32bit and lets face it 90+% of the machines gimp is going to run on are 32bit. I have GIMP 1.2.1 compiled and running on a Tru64 4.0f ES40 with 4 667cpus and 16GB of ram in it. On it the memory issues are even worse. I never bothered to try and get it working better since its not a practical solution anyhow.

Image redraw and processing. The image redraw with large data sets is slow. I have seen gimp 2.0 in action and it appears to be significantly faster than the 1.2-1.3 series. It is still long from being comparable to Photoshop 7.x.

Filters are just not as fast for the most part. Some are actually faster but a good example is the plain old sharpening mask with preview and including applying and redraw. It take about 5 times longer in GIMP on the same machine and file.

TIFF tags such as world coordinates, projection systems and such. GIMP should not trash this information. I could have swore at one time GIMP did not but it seems to do it again or now does it. Photoshop has always destroyed this information. I suppose this could very well require just a recompile of gimp with newer tiff libs or perhaps the geotiff libs(if that can be done).

I personally like the GIMP a lot. I use it with everything else in life but at work on large imagery it just isnt going to happen. It kicks butt for web graphics, home photo manip and art. It has a ways to go though to be on par with Photoshop 7.x and it makes sense. PS has been around for 4-5 times longer then GIMP has it not? They have had a lot more time to develope it and for the time frame in which gimp has developed it level of maturity is really quite amazing.

Ill put on my flame retardant gear just in case...

------------------------------------------------- This mail sent through IMP: http://horde.org/imp/

Hi,

I want to point out that we, the GIMP developers, are well aware of the problems you outlined. A few of them will need substantial changes and are supposed to be addressed with GEGL. Others (like TIFF tags) are rather trivial to fix and I wonder why you did not attempt to fix them yourself or get someone to fix them for you.

So much said, your comments don't fit very well into this thread and I changed the topic in an attempt to keep discussion on them separate from the funding discussion.

Sven

PS: I assume that you tuned your tile-cache setting and did not attempt to work on 500MB images with the default settings, or did you?

Quoting Sven Neumann :

Hi,

I want to point out that we, the GIMP developers, are well aware of the problems you outlined. A few of them will need substantial changes and are supposed to be addressed with GEGL. Others (like TIFF tags) are rather trivial to fix and I wonder why you did not attempt to fix them yourself or get someone to fix them for you.

It was not a problem in production so I saw no reason to address it. In the future though at least in this industry it definitely will become one for both applications and I am sure many others. They had also decided to move to PS further reducing the purpose of adding the support.

So much said, your comments don't fit very well into this thread and I changed the topic in an attempt to keep discussion on them separate from the funding discussion.

Sven

PS: I assume that you tuned your tile-cache setting and did not attempt to work on 500MB images with the default settings, or did you?

That I have.

------------------------------------------------- This mail sent through IMP: http://horde.org/imp/

On Feb 25, 2004, at 10:35 am, Mark Shuttleworth wrote:

Image manipulation is one of the key application areas that needs to be addressed for open source tools to become the mainstream desktop environment. I'm currently funding a number of different open source projects, and am considering funding work on the GIMP or CinePaint.

I can only speak for myself but being somewhat impressed by the GNOME bounty system, I'd like to recommend you the following to boot:

- Identify problem areas (talk to users, let usability experts speak up, etc.)
- Split up tasks in handleable subtasks - Set up a website specifying all taks and the possible money to get - Wait for people to show up and implement the tasks - Pay the money

The bounty is to be paid once the feature or solution has been successfully integrated into the main GIMP CVS repository and acknowledged by the maintainer(s).

This IMO works much better than any global funding where misc. people benefit
while others probably doing the more interesting jobs do not.

Servus, Daniel

Daniel Egger escreveu:

On Feb 25, 2004, at 10:35 am, Mark Shuttleworth wrote:

Image manipulation is one of the key application areas that needs to be addressed for open source tools to become the mainstream desktop environment. I'm currently funding a number of different open source projects, and am considering funding work on the GIMP or CinePaint.

I can only speak for myself but being somewhat impressed by the GNOME bounty system, I'd like to recommend you the following to boot:

- Identify problem areas (talk to users, let usability experts speak up, etc.)

Very well them,
I think that regardless Mr. Shuttlework would choose to act on these lines, this is exactly the "talk to users and usability experts" stage. :-)

So, let´s try to make an objective time line of the GIMP, like had been outlined before, and try and guess what could be sped up if there were apropriate funding?

The GIMP 2.0 is getting out AnytimeNow(tm). After that, GIMP 2.2 with a lot of enhancements, but no changes on the bit depths (but according to what I've heard from the GEGL guys, with a little GEGL already in) should follow in 6 months or so.

An them, in about one year, which IMHO is optimistic seeing the pace for the 2.0 release, there would be a GIMP 3.0 whith full support to other bit depths and color spaces, due to full integration with GEGL.

That is about it that has been said around here, ain't it?

Now what to say? these timelines apply if most of us go working on our spare-times, with a few exceptions. Maybe with funding, more people could go fulltime/partime into the project, and say, speed up GEGL development and integration. I think this is what we could think about and tell Mr. Shuttleworth.

Also, he asked reasons to support GIMP rather than, or concurrently with, CinePaint. IMHO the GIMP interface has evolved more and better than CinePaint's. But I use that program too litle to know about all of it's features - and their's todo list posted here seemed ratehr impressive.

Also, the inner code of the GIMP is expected to be cleaner and easier to extend than CinePaint's. But I had not actually picked into their code.

- Split up tasks in handleable subtasks - Set up a website specifying all taks and the possible money to get - Wait for people to show up and implement the tasks - Pay the money

Maybe mr. Shuttleworth would rather leave these steps to us - or to the Gimp Foundation(s).

The bounty is to be paid once the feature or solution has been successfully integrated into the main GIMP CVS repository and acknowledged by the maintainer(s).

This IMO works much better than any global funding where misc. people benefit
while others probably doing the more interesting jobs do not.

Servus, Daniel

Regards,

JS
->

Hi,

"Joao S. O. Bueno" writes:

The GIMP 2.0 is getting out AnytimeNow(tm). After that, GIMP 2.2 with a lot of enhancements, but no changes on the bit depths (but according to what I've heard from the GEGL guys, with a little GEGL already in) should follow in 6 months or so.

Yes, that's basically the short-term plan.

An them, in about one year, which IMHO is optimistic seeing the pace for the 2.0 release, there would be a GIMP 3.0 whith full support to other bit depths and color spaces, due to full integration with GEGL.

I would rather expect more 2.x releases than another major version jump with another 3 year development cycle. Now that the code is quite well organized, we should be able to carefully add new features without giving up stability. This will allow us to a more frequent stable releases. Our users will benefit from this since they get access to new features shortly after they are introduced. However, more detailed plans for the time after the 2.2 are supposed to be made at this years GIMP developers conference.

Sven