Press "Enter" to skip to content
  • mirror2image

    Ok here is your comment 🙂 . So how OpenFramework is better than OpenCV ?

    • bzerk

      OF is a C++ framework including a set of useful libraries, mostly oriented towards generative visualization, physical computing, etc. OpenCV is one of them. OpenFrameworks with OpenCV is better than Processing with JMyron or Processing’s OpenCV-derived libraries. I was just saying that I’m behind the curve in not using OpenCV. But machine vision isn’t what I’ve been focusing on. I just put Myron in to make a point. It just labels and numbers color blobs in boxes. It isn’t anything you can’t do by working directly with the pixel matrices, but like I said, not my focus. And I’m colorblind, so coding anything with color values confuses the hell out of me. And I haven’t made the jump to OF because I’m still building on the Processing sketch.

  • mirror2image

    So OF is just another OpenCV middleware layer. I was hoping someone rewritten OpenCV with better API. IMHO OpenCV API as it is now is butt ugly. cvCreate/cvRelease for such simple thing as matrix multiplication, no operator overload…

  • bzerk

    I’d hardly say that it is “just” a middleware layer for OpenCV. More like a middleware layer that wraps a lot of APIs, and is suited for a particular development niche. I’m afraid that your complaints with regards to OpenCV go over my head. I haven’t touched it yet myself, am using JMyron, and am using it for nearly nothing at that. Proof-of-concept to demonstrate the principle of inline machine vision fused with other AR types. I’m looking at the CxCore reference, but I’m not savvy enough to understand why allocating the memory structures to do the calculation is a problem. Is it something that should be more or less abstracted? Is it a memory management thing?

    And call me what you will, but “<<” stuff annoys me when I’m trying to think with OOP abstractions and pointers and such already. I’m a wuss when it comes to polymorphism so far. I’m no pro coder, homey.

    And I don’t really know or really care about the most efficient way of doing what I’m doing right now. For machine vision I’d need to, but I’m not even using image analysis in the current branch. Somebody will rewrite it when it needs to be rewritten. Just a hack to drive my hardware.

    Clearly, I’m not somebody qualified to pass judgement on the ugliness of the API. Check out the OF reference, and let me know if there is actually another layer in there. I was under the impression that the mappings were literal, but haven’t looked yet.

  • bzerk

    LOL 😀 You totally got me off on a code tangent when I was supposed to be coming up with a generalist’s narrative on AR. I just had to ask for comments. 😉 Would love to continue the conversation in a couple of days, since it isn’t one that I can have off-the-cuff and intelligently, and I’m the distractible sort. I’m happy that your comment was one that got my nose into another reference I’ve been avoiding. Thank you.