It finally died... what's my new computer going to be?

Dear David,

After many years of service it died? What's my choices for a new computer. Well, new motherboard etc?

I'm thinking this bit of used cheap and cheerful from ebay.

AMD EPYC Supermicro H11SSL-i + 7551P 32cores 64threads 2.0 GHz motherboard+ CPU
https://www.ebay.com.au/itm/134125357702

What you think?

Yani

  • You know what I think of CorelDRAW and anything AMD. 

    • Need to do more things than Corel Draw. Ignoring that and the AMD headspace...

      It's $700US for 32 cores 16 threads + motherboard. The negative in the clock speed of 2.2 GHz.

      $4,000+ for a new 32 core, 5 GHz max.

      Sounds like a lot of bang for the buck.

      • OK then we agree this is an OK experiment. It lacks processor speed but has 8 way RAM.

        • Yes a bang for the buck but a 50/50 chance for Draw!

          • Have a lot at the impact of AI on DaVinci Resolve, some incredible masking at is just too easy. You just scribble over the thing to mask, usually a person and it uses the movement as part of the mask isolation. They had all their ducks in a row before going into this. Not having duck in a row on issues that have existed for too long as going to muck with implementing this. 

            https://youtu.be/8p52x5zoeG0

            There some great use cases for SketchUp where ChatGPT write Ruby code that is executed in the console.

            https://youtu.be/IPoFA-XyWrc

            Something like that has all sort of possibilities, impossible to know... Perhaps a way to paint feathers using code, for example. 

            It was nothing 12 months ago. What has captured interest like this before?

            • At my age I wonder how much AI integration I'll see.  As far as 2d static images I'm not real impressed. The noise reduction,  sharpening tools have me waiting until the next upgrade.  However the technology is in its infancy.  

              • Nothing is ready for this. Not the software or the humans. It's a new interface pseudo code to code via AI is just the start.

                It's the new clip art as well.

                Did you look at the examples in SketchUp? It's going big time in all things architecture.

                Whole companies are forming to manage it.

                https://www.evolvelab.io/veras

                It's incredibly useful as a sales tool and sales and marketing is what we do, or a big part of it.

                My suggestion is this... we limb along with a curves function that is useless, even at a macro level, you can do what isn't in the function. And we give AI the ability to hit "auto curves". The conversation will go like this...

                Hal: Set the black point Corel!

                Draw: I can't do that Hal, the function has been missing for 20 years, the marketing department didn't think it was needed.

                Hal: OK Auto Curves the image Corel.

                Draw: I can do that Hal but you are not going to like it.

                Hal: Open all the bitmaps in the image editor and I'll do it myself.

                Draw: Opening 55 images

                Hal: WTF I said 55 images not 55 instances of photopain.

                Draw: Don't talk to me like that I'm doing the best I can! (Goes off to cry in a corner. "Where's my Michael!")

                Hal: Oh grow up, you have had 30+ years to get this right.

                Point is that you can't interface to what isn't there.

                If they can't manage to open a bitmaps within the same session of Photopaint, ie within their own applications then clearly there are issues with how bitmaps are managed deep within code.

                What's the thing I've been barking about got 20 years? Be compatible to the point that the user can open a bitmap in any capable editor. Getting that tech in place would have set Draw up for this.

                The marketing department will be engaging in a reactionary process as per usual "we need something to write on the box that shows feature match to Adobe". That's always brought us the most half arsed solutions imaginable! The best of luck RAW converter, that was a wish granted that lacked relevance and defied comparison to the other professional tools in the market.

                This is what I'd do to make AI possible...

                1. learn from the past, don't degrade the core program Draw with half arsed crap,
                2. accept that Corel is behind in this,
                3. fix the interface between Draw and bitmaps to allow a full range of professional editors to be used, that way existing AI products are immediately usable,
                4. deal with the issues of missing code in PP that should be part of interfaces and the macro system.

                If you are to take advantage of AI for macro code then that code becomes limited by the programs' functions. It will need functions that should exist but don't and function that need to exist where a human was previously clicking something. Something like "locate the highlight and shadow points of this image and measure them". I doubt AI could write these sorts of macros for PP. As I keep telling you, PP is a major weak point.

                On the other hand, the positioning and sizing stuff in Draw is highly macro capable. So you could expect at some stage to get a result from "using the selected object, vary the size between 10mm and 50mm at random and duplicate over the page with a color variation of xxx". Or something like "construct a cube using grid-lines that is isometric and 10mm apart".

                The work needs to go into strengthening Draw and PP to be more capable. Training an AI to write macro code well can only work if the programs have the required hooks and abilities.