AI Gigapixel issues + feature requests

I’m in charge of R&D for a large post-production facility. I’ve been testing AI Giga pixel for a couple of weeks for many different kind of jobs and would like to submit a list of issues that has strongly limited the use of this software in many of our workflows.
We’ve tested this software on many Mac machines (MacPro late 2013, MacPro 12core 2012, iMac Retina 5K both with AMD and Nvidia GPUs).

  1. no support for PSD files both in reading and saving. In today’s workflows PSD is very common, especially in complex layouts. This means that every PSD must be re-saved in TIFF before the enlargement, but, as the main scope of PSD is to have layers, we come to point #2.

  2. no support for PSD or TIFF files with transparency and/or more than one layer. Currently the only workaround is to duplicate the file, merge all the layers (loosing some blending modes) and save a PNG file. Very time consuming.

  3. (related to #2) saving a PNG file takes an unacceptable amount of time compared to TIFF. Many operators stopped the process as they thought the computer was freezed.

  4. bad support for CMYK files as it changes the color channels values in some extent. Sometimes is not evident, sometimes yes, especially in the shadows. Currently the only way to have 100% color fidelity is to work in RGB.

  5. no support for the file resolution parameter. There is no way to resize a file from say 300 ppi to 1,000 ppi. Currently the only way is to calculate the needed enlargement percentage and then bring back the dimensions to the original size after the process. Very time consuming especially on large files.

  6. no support for different enlargements in the same batch. If you want to save time, you must create groups of image enlargements and create the batch (i.e. 10 images @ 200%, 25 images at 350% and so on). This could be a nice checkbox option to add (all the same or different enlargements).

  7. crashes on some enlargements. It seems to affect only the biggest enlargements but it’s not clear the threshold beyond which it happens. We’ve tried to lower the original resolution, to reduce the enlargement size and some other moves but it did not work. Very annoying, because it happens just before saving the file after the process, so you must reprocess the file and hope for the best.

  8. on some images with a grainy texture (i.e. skin, metals, fabric and similar subjects) it creates some blurring limited only to certain square areas (like rectangulars) when the reduce noise option is turned on. This becomes more evident when the strong option is turned on but it’s not uniform. There is no way to predict how strong the effect will be, so very often you must process the file many times with all the different options and check the final result. Please add a preview window so we can decide if the result is going to be ok or not and process the file only once.

  9. there is no way to crop the image before the enlargement. This is very time consuming because we must create a cropped duplicate when in a layout the image is used as a portion of it and this happens rather often. Again, a preview window with a crop tool would be a great addition to the software.

Beyond that AI Gigapixel is a great software that produces amazing results and we’d like to increase its use in our working environment.

Kind regards,
Alessandro Bernardi

A very good list.

I am just a single at-home user, but personally I would like to see a way to call Gigapixel from the browser with a file selected. That is, on a Mac, be able to select an image, right-click and have Gigapixel start with that image. I do not see why that is not possible now.

I would also like to see the ability to call Gigapixel from a pixel editor like Photoshop.

Thank you for the feedback! I’ve forwarded it to the team.

bbmail, you can save a tif with layers. Take off the compression if it is on and tiff files do save with all layers intact. It is just a larger file than psd. Working in RGB is common. You only need to convert to CMYK for printing at the last stage, before sending the file to print.

Seabreeze, unfortunately AI Gigapixel doesn’t support TIFF with layers as I’ve mentioned before.
It can open TIFF with a single layer with transparency in some areas but the result is a flat TIFF with no layers.
The only way to have transparency is to use PNG though it’s better to save the resized file in TIFF because saving in PNG takes an enormous amount of time, unacceptable in a profesisonal environment.
As for RGB I’m perfectly aware that the CMYK conversion is a final step, we’re a prepress company.
Anyhow, we receive a lot of files already in CMYK from the clients and that’s the typical kind of file that must be enlarged ASAP.
So it takes a lot of time doing all these workarounds (merging all the layers into a single one, converting in RGB, enlarging, converting back in CMYK, changing the dimensions, etc.).

bbmail. I am very experienced in prepress and the production of printed material. Could you not get your clients to send through a flattened tiff? I realise it may and probably will arrive in CMYK - can AI Gigapixel deal with that?

Feature request: Write-support for longer file names. AI Gigapixel seems to read very long file names, but when it saves, it truncates the names so that “2018-09-21 11.02.31b Sunflowers.tiff” becomes “2018-09-21 11_output.tiff”.

This is not an OS limitation per se.

There may have been a programming decision that simplifies changing and writing the file name. (The new and old names have the same number of characters, and hence can have exactly the same memory locations, making alterations dead simple.)

Whatever the reason, it is extremely inconvenient for those of us who rely on information beyond the date and hour the image was made.

This looks like the ‘.’ is being stripped. Try it without the ‘.’

I don’t think this is a char limit

You’re right. I should have thought of that. Finding the dot/period was the way I parsed file names back in DOS 2.x days. I can replace dots with dashes, but it’s inconvenience to have to do this every time I use Gigapixel. Dots in dates is a pretty common practice.

Can this easily be fixed in the programme? The full file name, minus an Gigapixel-added suffix, is readily available, since the string with written to the screen once the file is loaded. Reading backwards to the first dot plus one more character will give you the position to add a suffix, won’t it?

Come to think of it, I’ve run into the same stripping problem with a few other programmes, too.


I’ll raise this as a feature request to the developers, but there are likely technical reasons the team didn’t parse for . characters. I don’t think it’s impossible, but the technologies we use building our tools might have been a factor in the current design.

Three cheers for this great new product, A.I. Gigapixel!

I would like to submit a feature request. It may have been mentioned before. It’s something missing which I depend on and is included in other similar software purchased (Genuine Fractals, Perfect Resize, PhotoZoom Pro).

In your Enlarge Options, please consider making it possible to keep the scale at 100% and have an editable field showing existing resolution where you’d be able to enter the desired output resolution.

Thank you.

What i tend to do is save the image as a jpg and a jpg of the mask and blow them both up and combine the results.