Search:

The algorithm is different for grayscale and black-and-white, so it would not really help to keep all pixels in 1 byte increments. Also, I think having 8 pixels in 1 bit reduces memory access (and...

Well, this image class is being used for a specialized purpose of computer vision. So, I am optimizing certain features for this purpose.

I am layering on top of .NET Bitmaps.

Hmm.. very interesting.

I think the iterators are a good idea.
I would wait until profiling to optimize but the problem is that I am kind of designing it now...

Sorry I have this habit of...

About the "micro optimization", I think it actually slows things down due to the possible extra memory access. Either way, I think the compiler can optimize it as best as possible.

About knowing...

Hmm.. thanks for the feedback.

I am really just calling getPixelDirect in most cases to either count the number of black pixels or to detect 2 or more contiguous black pixels. However, I am also...

Thanks for help.

I don't have a benchmark or profiler, but the getPixel is called millions of times.
Thinking back, I think there was something like hardware if-branch prediction, so I'll try to...

Hi,

So, I have an image class. However, pixels can be stored either in 1 bit per pixel or 8 bits per pixel (black-and-white vs grayscale). The pixels are stored in a pixel buffer of size...

thanks anyways

Thanks, but is there any open source/free solution? Better yet, a feature in msvc++?

Hi,

I was wondering if there was any way for msvc++ to automatically search my code for all public members that could be private without causing any compiler errors.
For example,

class hello...

What I mean is that the compiler has the option of choosing the more efficient method for the target processor, so I do not need to worry about it.

Correct me if I am wrong on this, but the...

Efficiency. By me specifying float, the compiler has a chance to upgrade to double if it deems it more efficient. It cannot do the opposite.

Okay, thanks!

Thanks!

Just to double check, is this platform independent?

Hi,

I was just wondering what is the best way for me to eliminate these warning messages:

1>(197) : warning C4305: 'initializing' : truncation from 'double' to 'float'
1>(197) : warning C4305:...

Thanks for your responses all....

I read this article and am now thinking that trying to optimize std::vector is not worth the time and code confusion :X

I'll probably try to optimize other...

Yeah, that was my original intention. Now if only STL defined some way to reserve a certain amount of space in a vector via stack it would be easier...

Sounds like a good idea thanks!

Thanks, but why is it not possible to use .push_back() with tr1::arrays? Like, do I need to manually keep track of the number of elements I store? Basically, I want to avoid dynamic allocation...

Hi,

I was wondering if there was any easy way to define a fixed size vector in C++.
Basically, I am trying to avoid the inefficiency of dynamic allocations without having to deal with handling...

Okay, thanks for telling me about Boost/TR1's tuples.

I guess since the comma operator is overloadable, this the best that can be done...

The compiler just needs to convert the return values implicitly into pointers passed as parameters.


Sorry, I meant that they should add multiple return values directly to C++'s syntax. Too bad...

Yeah, but Boost tuple kind of makes the other method with passing pointers more convenient...

Well, I guess if they don't want to add tuples to C++ they won't, but I don't see any problems with...

I am kind of annoyed that C++0x doesn't support multiple return values as such:

int, int GetValues(int x, int y)
{
return (x), (y);
}

int main()
{
int x;

Nevermind thanks for the link.