What Am I Doing Wrong When Combining 2 Images?

[puppychew]

Hi Folks
I need to install a Rolls Royce hood ornament on a smashed up car for an illustration.
I found a hood ornament and a car. When I insert he hood ornament image and convert it to a smart object, and reduce the size to fit the car, it is very pixelated.
Please see the attached 2 images. Any advise on how I can do this so the hood ornament looks better?
Thanks!

 

[JeffK]

From what I can see, you're pulling a lo-res image of hood ornament into another lo-res image of the car. In a sense, the palatalization is already there in the car image - you're now "applying it to the ornament.

Here's a section of a higher res image of a Rolls Royce ornament I found:



Here it is inserted into your car image:



Totally pixelated.

Here's that higher res ornament pasted into a higher res image of an old car:


Some pixelation (*corrected) but far better than the one above.

Here's your ornament pasted into the higher res image of the old care:



Not perfect but you can see the need for higher res image of the car to at least give you a somewhat decent result...

 

[puppychew]

Hi - what do you mean by palatalization? I thought that although a low quality image, reducing the size would not matter and also making it a smart object.

 

[JeffK]

Hi - what do you mean by palatalization? I thought that although a low quality image, reducing the size would not matter and also making it a smart object.

Weird spell-correct - proper term is "pixelation"...

 

[thebestcpu]

HI @puppychew
Smart Objects allow all adjustments / changes to the Smart Object to always refer back to the original data in the Smart Object. So if you resized the Smart Object to a very small image (few pixesl) and then resized back up to the original size, there is no data loss and the image is as good as when you started. So it is really helpful when one makes multiple changes to the Smart Object (resize, transformations, etc) the changes are all made to the original and not incrementally along the way. However, just resizing down once, will appear the same as far as pixelation as if you were working with a pixel Layer rather than a Smart Object.

One thing you can check is the default interpolation method set in Preferences in the General section. If you used nearest neighbor that would produce poor results for you purpose.

Best to start with better originals or upsize them with a program such as Topaz Photo AI or Topaz Gigapixel.

Hope this helps some
John Wheeler

 

[JeffK]

Just to experiment a bit -

Interpolation set to nearest neighbor per @thebestcpu / John's suggestion and ornaments converted to smart objects.
Original ornament on the left, higher res ornament on the right, original auto image:



Then I upsized the car image in Gigapixel, then dropped in the ornaments:



Better but pixelation still occurs. I've also noted that pushing Gigapixel too far and the image starts showing artifacts.

Here I went to the higher res auto image with the original ornament on the left and higher res ornament on the right:



I'm nowhere near the technician John is, but appears as if the ornament image is taking on the resolution characteristics of the car image.

Bottom line, without technical explanation, is to start with a higher resolution and sharper image.

 

[JeffK]

I've provided two upscaled images below created in Gigapixel - one is a 2X upscale, the other a 4X upscale. Maybe these will help.

2X


4X

 

[puppychew]

Wow - I got it now. Thank you sooo much!

 

[kabak]

Wow - I got it now. Thank you sooo much!

Your post in the thread What Am I Doing Wrong When Combining 2 Images? was deleted.
Reason: In the General PS Board,
you don't do the edit for the OF: you explain how it can be done.

In my opinion, it was superfluous to add how I did since I had to use a 3D sketchup file to adapt it to Jeff's image.
puppychew doesn't have that option anyway.
I offered the file for him to manipulate as he pleased in photoshop.