Klaus

Espresso, anyone? That's a clear case for caffè corretto. Lake Garda or wherever.

Even my E-PL5 uses some (simple) AI for face recognition. Subject tracking is for sure AI-supported in almost all cameras. But I don't think the intention was to exclude all AF-C images.

These rules seem practical and well balanced. You emphasize the intention (... substanitally altered ...) as the editing limit, rather than any specific tool or slider. 👍

Why not include a really tough category: No removal of ANY backscatter? Let's see who can place the strobes just right... This could be fun and perfectly honest. But at some point we have to acknowledge that we're no longer pulling the prints wet out of the developer. So as a general rule, I think it's too extreme.

The contest can (and should, probably) of course be starting next year or even later, no need to rush this. As for the rules - let's make that a community thing as well. "Excessive editing" is a purely subjective conclusion. This needs to be defined much more precisely to be a rule (which I find difficult). And "no AI" would downright exclude a categroy for smartphones / actioncams. Not sure we should go that way - but if that's what the community wants, these steps are easily removed from a standard workflow with a not too recent camera model. To me, this is not about perhaps winning a competitive event (there are enough contests out there for that) but to participate in a kind of fund-raiser within our small community. On this forum, we're discussing at length about equipment that costs thousands - so we should be able to cough up a few bucks from time to time to keep the site alive. "Featuring" someone's image(s) on the banner for money was not my intention, rather the notion that if we want to benefit from the site, we need to contribute as well. Just include an invitation that one should consider chipping in a small donation once in a while when uploading images for display. It seems that we're not paying with our privacy here and that's a fantastic thing!

I think that a member contest is a nice idea and paying a little bit per image also limits the work of the judges. I assume they will donate their time for this, so one shouldn't burden them with the entire content of the hard-drive anyways. Suggestion for a category: Freshwater - streams, lakes, quarries - anything close to home for the land-locked waterpixelers. I just discovered the "donation" item in the store. Why don't we say that whenever we upload an image for the member showcase banner (the one running on the main page) it's an honor to donate a little bit to the site for the display?

Since this is deviating even more from the sensor format - here's today's counters from the Forums page on Waterpixels: Photography Gear and Technique has 8.7 k posts. Video Gear and Technique has 1.1 k posts. Lights, Strobe, and Lightning Technique has 3.3 k posts. I consider these categories as gear-related. Together slightly more than 13 k posts. Photo/Video Showcase and Critique has 1.5 k posts. Shooting Technique, Workflow and Editing has 607 posts. One may consider these as a bit more "artsy". Together about 2.1 k posts. So, it's a clear yes, we are gear-heads. Yes, this is what you can do when you're dry: Shopping - or at least think about it, until you face your bank account. But >2k posts on more artistic topics is not too bad either. Plus there's the really nice running banner of pics on the homepage, not included in the count-up. We do have composition, appreciation and even discussions about conservation going on on the site as well.

🧐... we also need to provide a reason for the companies to sponsor our site! Such a discussion is, (un)fortunatly, only a couple of clicks away from the shopping basket. (and that is part of the fun)

Dear Fred, another thing you may want to consider for your grant money is a computer / workstation for image processing. I don't know if you are already well served for that at your institute, but image processing can become part of your routine - if you decide that it's appropriate. Using AI to improve image quality is certainly a difficult thing for scientific purposes. The borders of what's acceptable and not are blurry and also depend on the field of research (i.e. whether and how the community accepts these technologies). For cell biology, this is starting to become applied to improve signal/noise of e.g. fluorescence images, in particular before automated image analysis in "-omics" projects. But one has to be very vigilant that the "improvement" does not "iron out" true information or, in your case, introduce artifacts that you may count wrongly as a fish scale. Let your own judgment and gut feeling guide you as to how far YOU find this acceptable. Now, that said, AI processing can make counting scales a lot easier for you. I process my images with a program called ON1 photo raw, the also offer a Photoshop / Lightroom plugin for just the de-noising and sharpening. I wrote a tutorial about it on this site some time ago, certainly the tools from other software companies will be comparable. What can this do for you? Here's an out-of-camera jpg from last summer (Olympus E-PL5, 14-42 RII kit Zoom and Inon UWL-100 wide-lens without the dome): And here is the same image after processing: There's a bit more going on than just the sharpening / de-noising, but I hope the difference is visible even after the website has down-sampled this a bit more. The second picture will make it much easier to count the scales, but if you squint you can see them all in the camera-jpg as well. ON1 offer a free 30-day trial with full functionality (google on1 photo raw). You can try it out on photos from your current setup, see what this can do for you. It will work on both jpg and raw but obviously raw gives better results. If you try this out, you will see that you need a reasonable graphics card for your professional needs - otherwise it is way too slow. The GPU's are viciously expensive these days, so count ~3000 € or above for that workstation. This is a crop from the 16 Mpx M43rd sensor, the E-PL5 has the same sensor as the original Olympus E-M5 from 2012. Here's the full image (processed), limited artistic value but this is a technical discussion.

Sounds like you found your partner! The best results will come from equipment that you are familiar with - whether it’s is new or top-of-the-line does not matter if you can’t find the button for making a particular adjustment. I shoot a 12 year old m43 and it’s not the camera that limits my shots… As I mentioned, your « partner ». 😉 we keep those around for a while…

@Klaus For the strobe : my INON D200 (I also have 2 S2000) have a descent recycling time (1,5 sec). 20 GN is not a lot but it is usually enough for me. I read that popular strobes like Retra Pro Max II does not recycle faster. But I should investigate other brands. It is true that recycling time is limiting for me. This is why I have considered buying a powerful video light (like weefine solar pro) and test using burst mode photography, but fish seem to be very sensitive to such lights, so no sure this is a good option. Prehaps investigating a Backscatter Hybrid Flash HF-1 ? Sorry, my wording was imprecise. If you had a more powerful strobe, then it will not use all the capacity for the flash and you can get another shot very soon after that. The recycling could happen in the background, so to speak. I should thus have written … a stronger strobe with fast recycling … or so. And, as said before, in your situation you certainly want to rely on automatic TTL strobe exposure. The HF1 could be a good choice but I don’t own this strobe.

It's just a bit difficult to change a tele-converter during the dive when the 80 cm fish shows up... Scientific work should not be limited by equipment, I agree, but coming from a G5 with CMC2 the larger kits may be challenging to handle at first. I suppose that for macro / flat port setups all of the sensor options (MFT to FF) will end up not all that different in size. Probably it's when you use a dome port that the bulk for a FF system increases significantly? Let's not forget that the D200 strobe might be begging for an upgrade, too. Something that recycles really fast.

Whether the cropping is a benefit depends on how shy the fish are. You can count the scales the better the closer you are when the photo is taken. Fish from 1 cm to 80 cm seems to be a classic dilemma (and I have nothing to add to the suggestions). But it will always be better if you don't need to crop because you were close enough to fill the frame.

Noise & dynamic range: Maybe one should stop discussing sensor sizes and instead refer to pixel sizes? After all, you can buy full-frame cameras from 24 Mpx all the way up to 61 Mpx. The larger a pixel is, the more light it can collect and as far as I understand this translates to higher dynamic range and better signal/noise ratios. Considering that the full-frame sensor has a 3.8-fold larger surface than a M43 sensor, the pixel size of a 20 Mpx M43 sensor should correspond to a hypothetical 77 Mpx full-frame sensor. If those two sensors were built exactly alike, then their signal/noise and dynamic range should be the same (per pixel). I know there are other things to consider, but let's keep it simple. In that comparison, the 61 Mpx FF camera is thus not soooo different from a 20 Mpx MFT. Sure, you can add up neighboring pixels from the 61 Mpx shot and thereby down-sample to 20 Mpx. This will improve singal/noise, but then you lose the argument that the 61 Mpx allows for much more cropping. (Interesting to follow where this thread has gone from the original post. Special thanks for the history and archeology info, VERY much looking forward to more posts on this and especially to the eventual opening of the museum! Also the practical info is wise - you can pay a lot of extra luggage for the price of a new kit, I agree.)

There clearly are ways to reduce noise and INCREASE sharpness at the same time during post-processing. Yes, this is AI - at its best. And these algorithms are likely driving much of the improved signal/noise ratio in the new camera bodies' jpg engines (no matter which sensor format).