TonyB

It's only a problem until we ship the 1.6 Mac version in a few weeks time. The iPad version is 1.6 file format and the App Store Mac version in 1.5. The beta is 1.6 though.

Could you DropBox me one of your TIFF files with a white point so we can have a look.

If we load a TIFF then it does't contain any white point data so we can't show an absolute temperature value. This should work for RAW files though,

Are theses RAW files you want to adjust?

Doesn't white balance do the same thing for you?

It's more of a platform issue than anything else. I'm assuming you are on Windows so we will add scanning when the platform can support scanning on 64bit applications.

I also can't reproduce this. If I export to PNG and place the image on top of the original document and set the layer to difference I get nothing. This should show any difference.

Vector magic is cheaper than Illustrator and much better but you are correct that is does nothing else. I don't see this as a bad thing.

All good points. Just because we don't always respond doesn't mean we don't read the suggestions.

It looks like the images are not pixel aligned. It looks like you have pixel alignment switch off on the top toolbar. The default for Photo is on so I assume you switched it off for a reason,

Fixed in the next beta.

This is something we hope to improve over time. To expand a stroke we currently expand to hundreds of line segments and then try and fit curves back to the result. This can often work well but is shape and size dependant. I can't put a timeline on the improvement but it is definitely on our radar.

We are hoping to release 1.6 in late September.

32 fold oversampling would have to be done on the while scene to solve blending issues. If coverage calculation is 25% of our render time then doing it with 32 oversampled would still likely cause a 8 times slower render. Please remember the all coverages are calculated by the CPU so GPU can't help here. Also if we then do a colour render and blend with the 32bit coverage mask then the line contributions due to alpha would be degraded. In 3D this often doesn't mater but it it can make a bigger difference with possibly thousands of blended alpha primitives. High-end 3D renders moved over to raytracing for good reasons.

3D applications have the same issues with rendering multiple primitives. Your example of triangles is for a single primitive made of many triangles using the same colour or pixel shader. If it was simple to solve then we would have done it as would Adobe. We can always improve singe primitive rendering and we have plans to achieve this but it will not improve multiple primitive blending as your example. The workarounds for output are as you described - supersampling. We could add a supersample option on export to achieve this.