And again about crashes or the birth of AI

Perhaps every printer knows: bleeds are an eternal topic. An endless topic.

There is a lot of information on this topic on the Internet. With explanations, explanations, pictures and even videos. And it seems that even everyone understands everything perfectly, but when it comes to practice, we get the same picture: Business card, 90x50. Receive and sign. How does it not fit?

I understand, the Client is lazy. I'm like that myself. I didn't read the instructions. The result was, to put it mildly, unexpected... then I read the instructions and understood where I was wrong. But that's me. But my Client is not me and, accordingly, the approach, apparently, should be different.

So, the Client uploads the original layout to the site. The site tells him: “Sorry, we need 94x54.” Why? Who said? For what? The business card must be 90x50!

No. This is not a conspiracy of printing houses. These are technological requirements. The point is simple. Take a piece of paper and draw a square. Paint it over. And then try using scissors to cut along the edge. We bet it won't work out perfect? And so that there is no white along that very edge, you just have to cut according to the image. Just a little bit, even a hundredth part, but according to the image.

And here’s the question: isn’t it easier to immediately make the background image a little larger and crop as much as needed? Of course, it's easier. Congratulations, we just made those very bleeds.

But in practice the picture looks a little different. Without going into details, the Client fulfills the requirements of the site - increases the size of the canvas to 94x54 mm. In other words, it creates white edges. What do we get as a result? Of course, a complaint: “Your cut is crooked!” Of course... curve... the layout was even...

And here’s the question: what does Artificial Intelligence have to do with it?

Once, on the way to work, sleepily looking out the window and thinking about tired of meetings and neural networks, the thought arose: why not? I mean: others can persuade the computer to identify Teddy Bear with 80% probability...

And yes, I could. I persuaded my server.

Considering that this is, so to speak, a pilot project, the warnings are purely advisory in nature. I mean, the site recommends paying attention to bleeds: are there white edges around the perimeter? Oh, do you have a white background? Also good. Then, if we look from the position of “it happened by chance”, we will try to determine: are significant elements outside the safety zone?

Thank you for your attention,
Sergey Shulaev.