What's new
By:
Filters
Photoshop Gurus Forum

Welcome to Photoshop Gurus forum. Register a free account today to become a member! It's completely free. Once signed in, you'll enjoy an ad-free experience and be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

making smart objects smaller


P

puppychew

Power User
Messages
261
Likes
12
Hi

I am putting an exterior light fixture on a image of a house. I saved the image of the fixture from a catalog, made a good selection, then a smart object, and dragged it on to the house. It was a nice quality but when I transform-scaled it to 90% smaller the quality was awful when zooming in a little.

Is this how it is supposed to be or is there a better method.

Thanks
 
Did you finish the transform?
The layer will be of a lower quality whilst a transform is active, only once you commit to the transform will the quality return to normal.

Being a 'smart object' means you can scale up or down without loss of quality but there are limits, especially if the object is very small to begin with and you subsequently make it even smaller.....the less pixels the less detail....smart object or not.

Regards.
MrTom.
 
What is your image interpolation set to?

Preferences > General > image interpolation.

I tried Bicubic sharper and got better results when reducing smart objects.
 
actually does the bicubic sharper matter when reducing the image by transform? Maybe the original image should have been reduced before making a smart object. Tried that - no difference.
 
Last edited:
Yes, I tried that as well. I'm still searching............interesting issue.
 
How have you replicated the problem IamSam?

Regards.
MrTom.
 
Well, my smart objects do the same thing when reduced. I really didn't have to replicate it. I actually noticed it awhile back on another project but never checked into it.

This is a 1990 x 1440 image that I converted to a smart object.
I used Transform to reduce it's size by about 80%

The reduction looks great at 100%

Screen Shot 2014-07-08 at 7.57.16 PM.png

As soon as you zoom in it starts getting ugly...

Screen Shot 2014-07-08 at 8.00.28 PM.png
 
You weren't kidding, that IS ugly!

I'm not having this issue, which is why I wanted to confirm I was doing the same thing...namely convert layer to smart object, ctrl + T for transform, type in 80% for both width and height and accept.

smart_obj_01.png

Ugly free!

smart_obj_02.png

I'll have a dig around although nothing immediately springs to mind.

Regards.
MrTom.
 
Try typing 20%.

edit: Typing 80% for h and w is only a 20% reduction.
 
Last edited:
Well I guess this is what will happen if you reduce an image by that much, I didn't realize the OP reduced the image BY 90%, not TO 90% as is the way its indicated in PS.

An image of that size reduced to 20% is removing 2.3 million pixels from 2.8 million pixels.....its not really surprising that with the 570000 or so pixels that are left that the image is a little below par, in fact to be honest I think considering the amount of missing data PS does a pretty good job of it.

It sounds as if the OP's 'house' image is just way too small for its purpose and no amount of fiddling is going to help.

Regards.
MrTom.
 
The house image is 5000x2248. The light image is 400x400. But this image is just a light on a house and needs to be small. Yes it is reduced quite a bit. I just thought since it was a smart object there would be no problem. I guess I would be best using a light picture that is a smaller size to be sharper.

Also, just to make sure I am understanding this right (confusing) - If I want a 600 px width image on my website. Taking a photo that is 1000 px wide would show up sharper than taking a 5000 px wide photo and reducing it. Is this correct?
 
puppychew said:
The house image is 5000x2248. The light image is 400x400. But this image is just a light on a house and needs to be small. Yes it is reduced quite a bit. I just thought since it was a smart object there would be no problem. I guess I would be best using a light picture that is a smaller size to be sharper.
Click to expand...

So you are reducing a 400px square image BY 90%.....
Thats 160000px in total down to 16000px.
Keeping that square means its only 126x126px.....which is virtually nothing...especially in an image 5000x2248px.

It depends on what the image is for as to whether this is acceptable or not, I doubt this will be viewed digitally at 100%, (except by yourself), so being just a tiny proportion of the whole image shouldn't make that much difference.....it'll be so small its negligible.

You've done the right thing though, the bigger the images you can work with the better the end result.

puppychew said:
Also, just to make sure I am understanding this right (confusing) - If I want a 600 px width image on my website. Taking a photo that is 1000 px wide would show up sharper than taking a 5000 px wide photo and reducing it. Is this correct?
Click to expand...

LOL...yes it is confusing which is why its difficult to explain without making it worse.

Reducing, and more so enlarging images is considered the ultimate no-no, period.
Sometimes there just isn't any other way so if you have to do it keep the amount down to a minimum. If an alternative image nearer the right size can be found its best to use it rather than transforming up or down. Having said that it does depend a lot on the image content, some will reduce better than others....similar colours reduce better as the difference is closer to the original. High contrast images tend not to do so well as the difference is further away from the original.

Basically the less you mess about with something the closer it will resemble the original. So in answer to your question yes, all things equal, an untouched 1000px image would be better than a reduced 5000px image........unfortunately things are very rarely equal.

Regards.
MrTom.
 
I think that the key to understanding why the observation of pixelation when zooming in on a smart object is perceived as an unusual effect / problem can be found in these statements:

"...the quality was awful when zooming in a little...." - Post #1 by the OP

"... The reduction looks great at 100%. ... As soon as you zoom in it starts getting ugly..." - Post #9 by Sam

First, I think it's safe to say that anyone who has even a beginner's level of knowledge of PS expects ALL non-vector images to look pixelated if zoomed in past 100%.


The "problem" occurs because of a mis-perception of what smart objects are supposed to do and how they work. Smart objects only store the initial state of the object before any transform or other operation is applied to it, and they store the final state, after the transform. HOWEVER, THEY DO NOT KEEP RECOMPUTING THE FINAL STATE AS YOU CHANGE THE VIEWING MAGNIFICATION RATIO, as I suspect the OP had hoped.

For example, in the case being discussed in this thread, the smart object stores the full rez version of Sam's mountain, and it also stores the version after it had been shrunk to 20% of it's original size. When the user asks to view the result at some different magnification ratio, it takes the "after" version and magnifies or shrinks it in exactly the same way that it would display any pixel containing layer. If you zoom in substantially past 100%, it gets pixelated.

The benefit of smart objects is that if, say, after looking at the 20% result, you realize this is too small and you really should have only shrunk the object to 60% of its original size, not 20%, had you been using an ordinary, non-smart layer, you would have little choice other than to expand the 20% version by 3x in linear dimensions. This is guaranteed to produce a horrible result. In contrast, if you had been using a smart layer, when you change the parameters of the transform to produce a 60% version, it pulls up the "before anything happened", 100% version and computes the new result from it, not from the 20% version, an obviously much better approach. This is a tremendous benefit, especially if you wind up tweaking the size multiple times to see what gives you the best "fit".

HTH, and hopefully I correctly understood the effect causing the concern.

Tom
 
Tom Mann said:
First, I think it's safe to say that anyone who has even a beginner's level of knowledge of PS expects ALL non-vector images to look pixelated if zoomed in past 100%.
Click to expand...
That's a given Tom, I was restating the OP's observation in response to MrTom's question about replicating the problem. I knew it would pixelate, I just wasn't sure how to explain why. Thanks for you explanation. I don't think the benefits of smart objects were ever in question but thanks for explaining that in as well.
 
I knew you knew, Sam. ;)

I kept referring to the example you posted only because it was the clearest example given.

T
 
You must log in or register to reply here.

Back
Top