enter image description here

The white rectangle that contains the typography is a Smart Object that was skewed with Edit > Transform > Perspective.

The problem is that when the it's skewed and then resized (smaller) the text begin to look pixelate and it doesn't work very well, especially for letter with the oval, like e, r, etc.

Is there a way to make the text more sharp? I've already tried to change to Crisp, Sharp, Strong and Smooth but the result is the same.


Here the same Smart Object resized at two different sizes.

enter image description here

1 Answer 1


Try increasing the pixel count: if your image is 450px wide, double it to 900 and then downsample at the time of export.

This will give you more pixels to work with when describing the curves and diagonals.

As an aside, in computer graphics--especially games--you may have encountered anti aliasing methods such as "2x FSAA"

What this means is that the image is upsampled to (or more likely, rendered originally at) 2x the desired pixel dimensions and then downsampled for display. 4x would be 4x pixels count etc. FSAA is used to counter the exact issue you have in your question.


two smart objects: top one is 900px image w/300px type non-rasterized; bottom is 450px image w/150px type. Perspective applied to smart object (i.e. not applied in original placed file) using a recorded action.

enter image description here

the 900px wide image with type rasterized and perspective applied, exported at 50% size.same recorded action as above

test 900

the 450px wide image with type rasterized and perspective applied, exported at 100% size. (recorded action resulted in different transform (prob pixel count difference) some of the quality difference may be due to slightly different transform matrix).

enter image description here

  • Thanks. I've added an image as edit of the question. So the only solution is trying to double the pixel of original image? Commented Dec 5, 2013 at 18:51
  • See samples that I added
    – horatio
    Commented Dec 5, 2013 at 19:17
  • Note also that your type is ridiculously thin, so there will be a floor limit you hit where the calculated width of the line is smaller than a pixel. Also, the point is not "export at half size" but rather "render at twice size." Need moar pixels
    – horatio
    Commented Dec 5, 2013 at 19:18
  • Thanks. Sorry but being a novice designer I need more details about your useful answer. First of all, what do you mean with "900px image w/300px type non-rasterized"? By the way, what I'm doing right now is create a Text Layer in original file, type the text, convert it to Smart Object, then use Perspective transform tool. The problem is that after this I need to resize (smaller) the Smart Object and it loses the sharpness. What's the wrong step? Commented Dec 5, 2013 at 23:59
  • The file is 900 pixels wide, the type size was 300 pixels. Type is "type", but to run a perspective transform on it, it needs to be rendered as an image instead. This is called rasterization.
    – horatio
    Commented Dec 6, 2013 at 15:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.