Great to receive your comments!

[paul.s...@telenet.be]

Hi Everyone,

Great to receive your comments! I miss the times when FM was constantly experimenting and corresponding. It would be great to revive this!

I can understand that the blind enthusiasm for manipulating digital images has cooled, but it's unwarranted. FM was initially presented as a digital toy, with the idea of allowing anyone to easily create impressive filters. The logo of the gnome with the magic wand sums this up nicely. This approach was clearly over-optimistic and naive, because it quickly became clear that much more was needed for a decent filter and to let it work reliably.

So FM was further developed using a C++-based language. FM's capabilities quickly increased, but its operation and documentation became chaotic. FM ultimately became a toolbox with many possibilities, but a messy way to find the right material.

I am a photographer by training with an enduring passion for experimenting with images and developing concepts. Through FM, I learned to program and can bring my ideas to life. It's an addictive experience.

The impetus for reconnecting through the Filtermeister Mailing List is my desire to create an inspiring manual for creating more professionally oriented filters. I have a lot of interesting raw material I want to publish. I want to make my own selection from the FM Wiki and rework it into a coherent whole with more context.

What mainly limits FM is the basic interface that appears when the editor starts. I disabled this in an early stage and use my own design as a GUI. This neutralizes many problems and limitations. Developing an alternative GUI is quite complex to ensure conflict free code, but once created, you can directly enter the working core of a filter. This is usually only a fraction of all the text lines. So it's important to have a few solid, blank, basic filters that allow you to get started quickly.

FM is ideally suited for specific filter applications that aren't immediately commercially viable, such as graphic effects, image and color correction, masks, etc. Such filters are primarily intended to support the range of classic processing operations, but they can significantly enhance overall image quality. I can elaborate more on this in a future email.

I'd love to hear your feedback on this initiative and whether it can generate interest in supporting it.

Sincerely,

Paul

---

[Ladislav Ledvinka]

FM is not popular anymore not because of the lack of development of GUI capabilities, but because of lack of GPU acceleration implementation the most, and why it will never become popular? Simply because there are three options that I know of, OpenCL for GPU+CPU multitask, OpenGL and implementation of NVIDIA's SDK where all three are almost impossible to be integrated into FM's code by knowing that PS SDK is lacking of documentation and examples on developing filters, especially those that need to cooperate with mentioned accelerators.

Also, what you mention is totally old and out of time stuff, such as raindrops, rainbow, lens flare effects etc, and for the needed effects on image correction, without GPU acceleration, there is no chance anyone to get to know this software anymore. So, is FM dead? Technically yes! Can it be revived one day? Who knows, if there are developers eager to rise it from the dead, although knowing that today, AI is becoming more and more interesting for the developers, the chances for FM to be among the living again is very hard to almost impossible.

[paul.s...@telenet.be]

Hi Ladislav,

I'm aware of the lack of GPU acceleration in FM. The question is how accessible is the alternative programming language you're proposing?

Is it clearly documented with practical examples?
Are you already working with it yourself?


Regards,
Paul

---

---

Van: "Ladislav Ledvinka" <catapult...@gmail.com>
Aan: "FilterMeister Mailing List (FMML)" <filter...@googlegroups.com>
Verzonden: Dinsdag 19 augustus 2025 17:24:41
Onderwerp: [FMML2] Re: Great to receive your comments!

--
You received this message because you are subscribed to the Google Groups "FilterMeister Mailing List (FMML)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to filtermeiste...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/filtermeister/9a2c5b16-d193-4a1e-99f5-4cd1b94081c7n%40googlegroups.com.

[paul.s...@telenet.be]

Hi All, Thx for reply.

I'm aware that coding in FM can't compete with current technologies. What's most important to me is the ability to actually implement and test my own image processing concepts. The filters I've already created or are still developing are primarily for my own applications and to demonstrate specific image concepts. I don't aspire to commercial opportunities in this regard. A processing time under 3 seconds for a typical image of 12 MB is fine with me.

I'm not stuck with FM, but it's a tool I've learned to use over time. Of course, I'm interested in more efficient and faster software, but how much learning time should I estimate to be needed to get started right away?

- OpenCL for GPU + CPU multitasking or

- FM Visual Studio...

At first glance, building filters with this software seems quite experimental and requires a fair amount of background knowledge. Or am I mistaken? Is it realistic to master this coding in a relatively short time?

Is there a "Get Started" manual like the one for FM, with working examples, where we can add and modify own features? I would like to see more documentation on this, if existing...


Regards,
Paul

---

Van: "Kerry Payne" <payn...@gmail.com>
Aan: "paul simoens" <paul.s...@telenet.be>
Cc: "filtermeister" <filter...@googlegroups.com>
Verzonden: Donderdag 21 augustus 2025 01:37:07
Onderwerp: Re: [FMML2] Re: Great to receive your comments!

Hi All,

I opened up some old FML code the other day to scavenge a useful algorithm I had created/used in an old filter. However using it for modern filters just does not make sense to me.
I would highly recommend you use Harry's VS version if anyone is thinking about new filters, the limitations of FML are just too big for any real value.
It is easier to use Harry's version as a link into PS and use existing graphical packages like OpenCV for 90-95% of your filter requirements. Why build a custom Bilateral filter that can only run on the CPU when you have a highly optimised version capable of using Cuda or OpenCL already there? 
In my case It has been easier to just create stand alone versions and bypass PS altogether.

Like Harry, I am now using AI models as a significant part of my workflow and GPU acceleration is a must in 2025. The days of people being wowed by a brightness or contrast slider are gone. There are also so many free image enhancing tools online that even PS filters aren't really required anymore.
FML was a great product 15 years ago but time has moved on.

I am contemplating making a very niche Photoshop filter with Harry's version when I clear my current workload. I will post it if I succeed.

Keep coding and post whatever interesting projects you are working on!

Kind regards
Kerry

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/825242433.186977527.1755619451585.JavaMail.zimbra%40telenet.be.

[Harry G]

Hi guys,

Image editing with Photoshop might become outdated soon or at least in a few years. Check Google's Nano Banana, which seems able to perfectly edit images with just a prompt.

What are now plugins will likely be Loras for these image editing models, which you can train with image examples to produce certain effects or looks.

I trained a lot of "image effects" into my image generation model like lighting styles, skin color, face beautifier or lowkey/highkey effects. Also possible to do that as "plugged-in" loras.

Cheers,
Harry

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/981574102.195313639.1755781955091.JavaMail.zimbra%40telenet.be.

[Harry G]

Hi,

Fully AI generated images have no copyright currently. For AI modified images the old copyright will likely transfer. If you modify an AI image yourself enough, you should have the copyright too. But a lot of lawsuits will still have to clarify this.

I generate AI images locally mostly. You can run smaller to medium models locally. Even quantized versions of bigger ones. 8 or 12 GB VRAM is standard, best 24 or more.

Lawyers & judges will ultimately be less needed because of AI either. If you lose a job because of AI, you should learn to do work with AI. Programmers will likely become project managers that tell AI what to do. Software will be written faster, but there will likely also be more demand. Technological advance just changes jobs & does not destroy them.

Harry

Jude Suszko <h6zaw1r7...@runbox.com> schrieb am Do., 21. Aug. 2025, 15:49:

Hello Everyone,

I suspect there's going to be a lot of legal issues regarding copyright ownership of images that were created or modified by AI.

I also think that a lot of AI stuff doesn't really run on your own computer because most personal computers can't run most AI models.   I think the computing is actually done "in the cloud", which might be a problem with images of a sensitive or proprietary nature.

My prediction is that all the people who can't find programming work because of AI will end up becoming lawyers.

Jude

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/CAJgfx9LGRTJGVGZbhcvo3QxMUbMwx1k8a9TURsthBtdx_TzM%3Dw%40mail.gmail.com.

[Richard Rosenman]

Hi guys;

 

I too am still on the list but have long ago stopped using FM. It was a nice tool for me to learn on but unfortunately without support it is vaporware.

 

I long since switched to developing plugins for After Effects (with Visual Studio C++), which supports animation and produces much more interesting results. You can see some of my plugins here:

 

DOF PRO: https://richardrosenman.com/shop/dof-pro-ae/ 

Particle Projection: https://richardrosenman.com/shop/particle-projection/

IK Studio: https://richardrosenman.com/shop/ik-studio/

Motion Mosaic: https://richardrosenman.com/shop/motion-mosaic/

Path Animator: https://richardrosenman.com/shop/path-animator/

Box Fitting: https://richardrosenman.com/shop/box-fitting-ae/

Palette Mapper: https://richardrosenman.com/shop/palette-mapper/

 

For the last 2 years I've also been developing an extremely powerful CUDA-accelerated full-featured particle plugin for After Effects. It is really coming along and, as Harry pointed out, it harnesses the power of the GPU which is really now the only way to develop software.

 

Harry, I too have been working with ComfyUI. This is the future for localized AI generated content. I have 4 GTX 1080 Ti's in my machine but unfortunately ComfyUI isn't multi-GPU so I need to upgrade in order to produce any kind of AI video with it. Therefore I'm saving up for the RTX 5090. ComfyUI is awesome but extremely complex and not particularly artist-friendly. However, I'm understanding it better and better each day and the capabilities are endless. It is becoming THE tool pretty quickly.

 

It all comes down to which model you use with it and apparently the new Wan 2.2 is the definitive model for home-grown AI video, so I have to try that out as well.

 

Adobe stock has been tanking for the last two years because analysts don't see it developing AI tools quickly enough and, to be frank, it needs to. The traditional software tools it provides are quickly being replaced with much of the AI tools coming out. It's own suite of AI tools, called Firefly, are ok but in my opinion, still not comparable to the competition.

 

It's an exciting (and somewhat scary) new era with AI. It will challenge and radically change many industries forever but will also provide new roles.

 

Harry, you should send me links to your work with the LLM's so I can check them out.

 

Regards,

-Richard

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/CAJgfx9LGRTJGVGZbhcvo3QxMUbMwx1k8a9TURsthBtdx_TzM%3Dw%40mail.gmail.com.

 

Richard Rosenman // Creative Director

Hatch Studios Ltd. // 1.416.565.6609

ric...@hatchstudios.com

http://hatchstudios.com

[Ladislav Ledvinka]

I am not proposing anything here. Besides, there is no alternative language here except C/C++. Not even thinking to continue the support of this software anyhow because it has a lot of flaws in it, from the concept of the integrated programming language to the concept of multitasking and GPU acceleration(not implemented at all) and AI integration(nowadays), just losing time to firstly understand what is going on inside the FM's code, and then to make the fixes and the integrations in it. The idea was very good at the late 90's or early 00's, but not now.

[Kerry Payne]

Hi All,

I opened up some old FML code the other day to scavenge a useful algorithm I had created/used in an old filter. However using it for modern filters just does not make sense to me.
I would highly recommend you use Harry's VS version if anyone is thinking about new filters, the limitations of FML are just too big for any real value.
It is easier to use Harry's version as a link into PS and use existing graphical packages like OpenCV for 90-95% of your filter requirements. Why build a custom Bilateral filter that can only run on the CPU when you have a highly optimised version capable of using Cuda or OpenCL already there? 
In my case It has been easier to just create stand alone versions and bypass PS altogether.

Like Harry, I am now using AI models as a significant part of my workflow and GPU acceleration is a must in 2025. The days of people being wowed by a brightness or contrast slider are gone. There are also so many free image enhancing tools online that even PS filters aren't really required anymore.
FML was a great product 15 years ago but time has moved on.

I am contemplating making a very niche Photoshop filter with Harry's version when I clear my current workload. I will post it if I succeed.

Keep coding and post whatever interesting projects you are working on!

Kind regards
Kerry

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/825242433.186977527.1755619451585.JavaMail.zimbra%40telenet.be.

[Harry G]

Hi Richard,

Check out https://github.com/pollockjj/ComfyUI-MultiGPU 

And https://github.com/neuratech-ai/ComfyUI-MultiGPU

Harry

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/587967849.27968.1755785982861%40email.ionos.com.

[Roberto Muscia]

Yeah, yeah: one day a big shiny new Pick-up truck everybody could afford/ride was received with open arms, opened a world of possibilities and was used for anything people could image up to a certain limit and a week later the very same thing is considered totally unusable trash because it doesn't have todays rear camera's, auto parking feature, self-inflating tires and comfy massage seats. Indeed a totally unusable vehicle...

• FM was -and still is- an understandable tool allowing non-IT-graduates to create a vast diversity of image effects or image data analysis.

• Did they realize it back then? Yes!

• Did they appreciate it back then? Yes, big time!

• Did they use it back then? Yes, even people who FM was not intended for: professional developers who created filters for a living!

• Were end-users of FM-created filters happy with the filters? Yes, big time!

• Does FM still offer the same features? Yes, and more.

• Is FM nowadays offering all features commercial developers 'need' or rather: want to use in? No, it misses modern technology support.

• Have commercial developers in the mean time ran into FM limitations, gathered more programming skills and moved on to more difficult and powerful development platforms? Yes, and they should... not stick to a laymen's tool for developing image effecs.

• Has FM therefore lost all of its affordable and easy to use features and turned into garbage? No, of course NOT, however...

• Is the average layman nowadays still interested in sacrifice for result? No, the very majority of people in a rich spoiled society where everything is taken care of prefer easily available instant fun and amusement instead: the layman is lazy.

There is no reason to call FM outdated crap. It still features what it did.

If you need professional features get yourself educated and buy modern professional tools.

Laymen, students or engineers still can use FM as a tool for realizing many of their goals and if they run into FM limitations they could go with plan B which is workarounds. It has proven effective in the past.

However, FM interest of photographers is probably what has shifted mostly: from major to non-existent anymore with todays huge availability in photo effects.

Students and engineers probably already have some other programming skills in their backpack they might be able to use instead of learning FM.

For potential and current FM users the only crap that IS fact is that there is no place for FM support anymore; The FMML was the only solid alternative for the outdated documentation while people now asking a question on this list get silence as an answer.

However, FMML is hardly ever visited by new people so that definitely proves a point.

Roberto

Op Tue, 19 Aug 2025 16:15:07 +0200 schreef paul.simoens via FilterMeister Mailing List (FMML) <filter...@googlegroups.com>:

--

Gemaakt met Opera's e-mailprogramma: http://www.opera.com/mail/

[Lance Otis]

Digital artists: Re all this discussion about FM being out of date. No question it is, but it still has uses. Here are a couple of examples using a 16 bit plugin for Irfanview to simulate Chuck Close paintings using reference photos and about 1100 drawing option combinations:

Best Wishes...

// Lance Otis

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/CAJgfx9LGRTJGVGZbhcvo3QxMUbMwx1k8a9TURsthBtdx_TzM%3Dw%40mail.gmail.com.

[paul.s...@telenet.be]

Hi Lance,

If I understand it correctly, you're suggesting these results were created with a plugin built in FM.
Could you tell something more about this? Is this plugin only for IrfanView, or does it also work in Photoshop?


Regards, Paul

---

---

Van: "Lance Otis" <loti...@gmail.com>
Aan: "filtermeister" <filter...@googlegroups.com>
Verzonden: Woensdag 18 februari 2026 07:53:02

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/f6382e3d-32a0-427b-ad82-60933a3b90b2%40gmail.com.

[Roberto Muscia]

Hi Paul,

Lance wrote it was done with - a 16 bit plugin for Irfanview -

Browsing through the list of available plugins for Irfanview I noticed only one plugin that was paint-related: the plugin is called 'Paint': 

• PAINT - (version 0.4.13.65): allows IrfanView to to paint lines, circles, arrows, straighten image etc.

Roberto

Op Thu, 19 Feb 2026 18:24:47 +0100 schreef paul.simoens via FilterMeister Mailing List (FMML) <filter...@googlegroups.com>:

[Lance Otis]

Paul, and Roberto: 

Found my 32 bit version 8BF. It works with Photoshop and Irfanview. Here is the List I loaded for Irfanview:

To view this discussion visit https://groups.google.com/d/msgid/filtermeister/op.3k17mmmefxw4o6%40roberto-asus-g11.home.

[Roberto Muscia]

Hi Lance,

It looks like the screendump is from an Irfanview plugin that allows to load and run 8bf files; probably:

PSHOST - (version 4.68): allows IrfanView to load Adobe Photoshop 8BF filters 

For that painting effect in the Irfanview plugin you 'loaded' a number of real 8bf filters and one of those is the Chuck Toolkit plugin.

I would think you have the 'Chuck Toolkit' Photoshop/8bf plugin installed somewhere on your computer.

However, I cannot find that particular plugin online.

On the other hand the majority of the plugins in your list are called '...Toolkit'; I can't remember adding that word to the name of a plugin was ever common. At least since I started to learn about plugins and if it was in the FilterFactory era I doubt decent Chuck painting effects were possible back then.

Roberto

Op Sat, 21 Feb 2026 03:57:37 +0100 schreef Lance Otis <loti...@gmail.com>: