1

I have photos and videos that are too low resolution, so they look awful when enlarged, although they are quite good in their original resolution.

Therefore, I would like to recommend free software (not online) that could increase it with minimal loss of quality, please?

Thank you

3
  • are these files that started as a digital file? or are these scans of a physical media like film or a printed picture? Also, were there any online services that you saw that could do this? Would be useful for trying to find similar offline services Commented Jun 2, 2021 at 18:35
  • @DeveloperACE These are photos and videos from a very old digital camera. However, sometimes I find a nice image on the internet that I could use for something, but it has too low a resolution. I searched the internet for how to do it with a picture, and it found me a bunch of online services and photoshop. Therefore, I would recommend some free offline SW. Commented Jun 2, 2021 at 18:46
  • @MichalRama, forget. No software can do them bigger and good looking. There is no information to fill the frames. Commented Jun 2, 2021 at 19:18

1 Answer 1

2

Taking a small image (less pixels) and scaling it up (to an image with more pixels) requires adding more information to the image.

Your options for doing this depend on whether or not this information is available and how you plan to fill in this information (which can depend on where you got the image).

From the internet

If you found the image on the internet, you can try reverse-image searching it on google (using the camera icon on images.google.com) and then filtering the results by image size. Oftentimes, you will find a larger copy of the image.

From Film

If the image was originally captured on film (and you still have the original film and/or negatives) then its possible to have it digitized (either by a dedicated service, or a more DIY method). The results of this will largely depend on the quality of the camera or other device used to capture the digital copy of the film image since film is essentially chemicals on some plastic and doesn't exactly conform to the modern digital notion of a "pixel". There is a limit to how small the particles/chemical compounds that make up the film can get so at a certain point, increasing the digital camera resolution wont help recover any more information since there is no more information that was stored on the film

From a printout, or other hardcopy

Physical printouts are similar to film in that you can "digitize them" by using the scanner that is likely part of your homes printer. Depending on the maximum resolution (measured in dpi, or dots per inch) that your printer can scan, you will likely be able to scan an image at a high resolution to capture more of the details in the ink patterns/information in the printed image. Just like film, there will be a certain point where the scan cannot capture any more detail from the printout because there is a limit to how many ink particles can be crammed into a piece of paper.

Digital

If the image was taken digitally, then your options are quite limited. If the image was taken at a higher resolution, then you can just make a copy of that higher resolution file. However, when scaling digital images down, oftentimes information is thrown away, making it next to impossible to scale it back up again unless you have the original image saved somewhere.

The reason I say next to impossible, and not impossible, is because, with enough time and energy, you MAY be able to use Photoshop and memory, or maybe some future AI upscaling technology to make intelligent guesses at what that lost information could have been. However, if you're looking for something that's free and available for download, this is very likely not going to be an option.

Upscaling

I also want to mention this as it may technically do what you need, even though it will likely not give you the result you are hoping for.

One other way that you can fill in the extra pixels to scale an image up is to make copies of the pixels that are already in the image. For example, if you want to take a 100x100 pixel image and make it 200x200, you can just take the color of each pixel and copy them, so for example the single pixel in the upper left corner would become a square consisting of four pixels (of the same color) in the upper left corner. this would have the affect of making your image twice as wide, twice as tall, and contain four times as many pixels, however, the image would still appear blocky and pixelated if you were to look at it because you are effectively making each pixel bigger while also making the image bigger.

Essentially, in order to scale up an image, you need to ask the question, "where am i going to get the extra information/detail/pixels from". Sometimes this is available fairly easily (the Internet and/or a digital photo taken at a high resolution and saved elsewhere). Sometimes, you have to re-scan some piece of physical media (Film, Printout) at a higher resolution to try and capture more of the information that is already there. You can copy the information/pixels that you have (Upscaling), but most of the time all you can really do is either make it up/guess (Photoshop + Time) or just accept that the additional information has been lost and live with your image as-is (and youll probably also want to protect it to prevent any more of it from being lost)

I realize this probably isn't the answer you were expecting, but unfortunately, the way that movies and TV portray "enhancing an image" (often multiple times) is not something that is realistic (at least in a way that's widely available as free/gratis software today.

4
  • This answer really sums it up, software cannot make high quality pictures from low quality pictures. Information which is not available in the original picture would have to be invented/guessed (by an AI) and this can or cannot improve the look of the picture, but it also makes it wrong. Commented Jun 15, 2021 at 9:00
  • yeah. i should probably point out that this kind of AI upscaling tech exists in the video game space in the form of Nvidias Deep Learning Supersampling (DLSS) and AMD's new one too, but those models are specifically trained for video games so that is likely not going to work for your average family photo Commented Jun 16, 2021 at 15:16
  • Ha, just like in the movies, where the AI can reconstruct the number A 235 547 from a single grey pixel 😲 (wonder how they trained this). Commented Jun 16, 2021 at 15:30
  • i dont think technology will be at that level for a while, but maybe google is already working on it for advertising purposes haha Commented Jun 16, 2021 at 15:44

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.