Source Image does not match the Preview

[Neutronman]

Howdy all,

Well, I decided to take the plunge and learn PixInsight.  Although I am a 20 year Photoshop junkie, I have REALLY been having fun with PI -- very impressive tool.

I have processed a couple of images and have had great results, but now I encounter my first problem...

I am processing the Iris and trying to apply HDRWT to the core area.  I have created a luminance mask to protect all areas except the core, created a preview, and apply the HDRWT to the preview.  It looks great.  I then apply the HDRWT to the full image, but it looks NOTHING like the preview.  In fact, it kinda looks the opposite (dark and muted instead bright and contrasty).  Any clues as to what I am doing wrong?

Thanks!

JOhn

[Carlos Milovic]

Which size is the preview? You should use HDRWT with full sized previews, because due to the multiscale nature of the algorithm, results will vary a lot if you apply it only to a small sample.

[Neutronman]

Understood, okay I'll try not using a preview.  My preview was 870x720 (source file is 3000x2000).

Thanks,

JOhn

[Andres.Pozo]

Understood, okay I'll try not using a preview.  My preview was 870x720 (source file is 3000x2000).

Thanks,

JOhn

Using a preview is OK, but it must the same size that the original image. The easiest way to create a full size preview is clicking in the tab and dragging it to the gray area under it. Please, watch the screenshot.
You can also drag and drop previews between different images or dragging a preview to the desktop of Pixinsight to create a new image from a preview.

[mmirot]

Understood, okay I'll try not using a preview.  My preview was 870x720 (source file is 3000x2000).

Thanks,

JOhn

Using a preview is OK, but it must the same size that the original image. The easiest way to create a full size preview is clicking in the tab and dragging it to the gray area under it. Please, watch the screenshot.
You can also drag and drop previews between different images or dragging a preview to the desktop of Pixinsight to create a new image from a preview.

I might add to this.
Note: It won't process any faster but still vary helpful if you want to compare results with original

[Neutronman]

Understood, okay I'll try not using a preview.  My preview was 870x720 (source file is 3000x2000).

Thanks,

JOhn

Using a preview is OK, but it must the same size that the original image. The easiest way to create a full size preview is clicking in the tab and dragging it to the gray area under it. Please, watch the screenshot.
You can also drag and drop previews between different images or dragging a preview to the desktop of Pixinsight to create a new image from a preview.

I might add to this.
Note: It won't process any faster but still vary helpful if you want to compare results with original

Yeah, that was the next question -- is a full size preview any faster?  I did not think so.  Too bad there is not a way to scale the process to adapt from a small preview to the large source file, since the HDRWT can take a looong time.  It makes tweaking a very slow process.  But the effect is great.

[Carlos Milovic]

No, the process itself will last the same amount of time. But, previews are stored in RAM, while undo states of the images are stored in physical memory. The result, is that previews will be faster in undo/redo operations, making a faster blinking effect. Also, you'd may use several previews, and compare results between them very easy/fast.

If you want to try it faster, you may clone the image, and integer downsample it by any factor 2^n (2, 4, 8, etc). n will tell you the difference in wavelet layers. Of course, this will be just an approximation, but it will be better than sampling a small portion of the image.

[Nocturnal]

I've actually had pretty good luck using sub-image previews to estimate what HDRWT would look like. Perhaps it depends on the contents and the number of layers you work with. Larger previews work better than small ones.