Hacker News new | past | comments | ask | show | jobs | submit login

Makes sense to me, though I've noticed that it's surprisingly hard to find a good alternative test image. This one has:

- Fine directional texture (the weave of the hat, hair)

- Fine non-directed texture (the feathers)

- Fine detail (eye glint, lashes)

- Medium texture (reflected hair)

- Sharp edges

- Curved edges (mirror frame, shoulder)

- Smooth gradients (skin, blurred beams on left)

- A closeup face

- Color

- Strong contrasts

I'd love to see a consensus around a good alternative high-resolution test image that has a bit of all of that (and maybe with a Macbeth chart thrown in somewhere too) with appropriate licensing, but none of the suggested alternatives (e.g., cameraman, mandrill) that I've seen quite have everything or are of any better resolution.

One source that I have recently been using for test images for my experiments, though, is the Wikimedia Commons Picture of the Year finalists collection: https://commons.wikimedia.org/wiki/Commons:Picture_of_the_Ye...




Specific modern technical issues with the original scan are:

    Lena contains about 250,000 pixels, some 32 times smaller than a picture snapped with an iPhone 6. And then there’s a quality problem: The most commonly used version of the image is a scan of a printed page. The printing process doesn’t produce a continuous image, but rather a series of dots that trick your eye into seeing continuous tones and colors. Those dots, Acton says, mean that the scanned Lena image isn’t comparable to photos produced by modern digital cameras.
and:

    [..] Scott Acton, the current editor of IEEE Transactions [..] and two former editors revived the old call for a ban, and proposed to the journal’s editorial board that IEEE Transactions institute a moratorium on Lena research. “In 2016,” he says, “demonstrating that something works on Lena isn’t really demonstrating that the technology works.”
~ https://www.theatlantic.com/technology/archive/2016/02/lena-...

the implication being that too many algorithms were trained on the Lenna image, to the point where the ASCII phrase LENA.JPG can now be expanded to the full original uncompressed image.

It, along with the Mandrill image, are quaint historic footnotes examples of types of images (drum scanned from magazine prints) that are no longer anything close to mainstream examples suitable for reference.

> it's surprisingly hard to find a good alternative test image.

https://arxiv.org/pdf/1202.6429v3.pdf

uses a licenced image of Italian heartthrob Fabio Lanzoni with skin tones and fine detailed blond hair.


Yes, putting aside the controversy, I agree that its sorely out of date and not very representative of modern images. I remember being amused when first seeing the paper with Fabio, but sadly it too is greyscale and of poor resolution.

Regarding many images being trained, I have mixed feelings. I think that the modern approach of having a solid corpus of images is important for quantitative analysis, but it's also nice to have a well-known common image to eyeball for a quick qualitative check.


I would add the consent of the subject of the photo. Indeed i think it could be a nice way of honoring a pioneer of the field by taking a photo of them that specifically contains the above elements.


Yes, I was thinking of consent as part of the "appropriate licensing".


Color and contrast are not ideal. There's no pure white or deep black nowhere (shadows are gray). And color... this just contains a dull blue and orange'ish.

I like the kingfisher from your wiki link, winner of 2020. Maybe we need more than one image for tests: one with perfect contrast range, one for human skin, etc. Instead of looking for one perfect image.


Couldn't you make a bigger image that has all of these?


True! It would definitely be a chance to improve.


Brazilian plastic artist Ila Fox created one alternative image: https://it.m.wikipedia.org/wiki/File:ILena.jpg CC-SA Licensed, original from https://blog.ricbit.com/2009/11/lena-e-ilena.html?m=1


Tbh insofar as the problem is the specific image used - recreating the shot with a reenactor doesn’t fix the problems with the model wanting their likeness to not be used for this purpose (and probably with the way some states are going could be considered protected itself, eg the Elvis act in Tennessee), and even opens up the argument that this itself is re-victimization and harassment (eg analogous to CSAM image arguments).

It also fundamentally doesn’t fix the problem with it just being kind of a horny selection of an image in the first place, which is the other half of the complaint.

I’m not saying it’s something that keeps me up at night personally, other than it just being generally a little tasteless. But for those who do argue against the use of Lena, it really satisfies none of the arguments other than the purest copyright one (and it may not even be legal to use in all jurisdictions because of personal likeness concerns anyway).

If you’re gonna get new material, get new material. Not just Legally Distinct Lana.


Oh, that one's interesting. I hadn't seen that alternative before.


Unlike in its time, almost everyone now has access to a good camera. Most of the effort would go to good lighting and finding objects that satisfy these criteria. A contrived CGI render could also serve the purpose.


Many colour-correction swatches use a pale woman with a fruit hat due to the colour range, gradual skin tones, and high dynamic range. I'm surprised it hasn't become the replacement standard image...


I am just making a wild suggestion here. May be ask AI to generate one with the above prompt?


If you want your image processing algorithm to be optimized for the real world you should test against real photos. It’s not like we have a shortage of photos so synthetic data is absolutely necessary


Give me ten k and I'll make the AI take the prompt and return the best stock image, guaranteed

The optimal lenna replacer, for boomers who think they can reasonably belabor this point

Just pay me up front, papi needs a quick cash fix scratches their neck and tries to swivel their head 360 degrees


No need, we have an ethically sourced Lena image here https://mortenhannemose.github.io/lena/


This is clearly a sly reference to an exclusionary in-joke, which makes it an in-joke itself, and thus also exclusionary.

Needless to say, I'm still offended. And I think we can all agree that no one should ever be offended.


I'm offended that you're offended. And I think we can all agree that no one should ever be offended.

Hint: this path or yours? It doesn't work. Feeling offended and asking others not to offend you has to be reasonable within society's moral values. They've changed, obviously. But there is a large group of conservatives (as in: the type who wants traditional values) who loathe that change. The group is significant enough that they can push back changes like these. Case in point: Trump got elected and ensured SCOTUS is now 6-3 conservative-progressive.


I am offended you're offended they're offended. And I believe the best defense against trolling is direct offense

Hint: this little old glory jig of yours? You know why you're offended? Because you have no principles or values worth fighting for. But you want to fight for something. So you fight for an inglorious past. And you can't think of a better way than to skirmish with sjws at random.

We see you, and you are transparent to us. If you were above the struggle, you would say nothing. Yet you want to fight. You NEED to fight. We all do. And yet you have nothing worthwhile to fight for. So you fight us.

The problem with your approach is that your way of thinking will die, since it produces nothing new or worthwhile. To the degree that your views get presidence, there will be misery and gnashing of teeth, stagnation, and eventual failure of the empire built on your non-principles. But the cause of true justice will remain living, until victory.


LLM gibberish.


False. The best LLMs speak with more empathy than I do

If I was using an LLM, I wouldn't be beefing with you. I'd find some way to engage you while respecting your point of view.

Nope. I'm definitely a full 100% grade A human a*hole


That's not ethically sourced Lena. That's mansplained Lena. And it doesn't add anything new or better.

Srsly, just hand over the 10k. I'll get the job done! Then we'll have thousands of Lena alternatives! And they'll be objectively better!




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: