Jump to content

Flea Yatsenko

Resident
  • Posts

    478
  • Joined

  • Last visited

Everything posted by Flea Yatsenko

  1. I hope they are realizing that the biggest issue isn't even search quality being poor, but SLMP failing to work sometimes. Random 404 and 5xx errors all the time is not good for an ecommerce marketplace.
  2. Getting 404 errors when uploading images to new listings and saving a new listing sometimes doesn't actually work. Every time I make new listings I get reminded how borked this thing is and how much lost income it's costing me because it doesn't work properly. Wouldn't be surprised if check outs and other parts of SLMP behaved the same way.
  3. You are comparing year over year, right? Like February 2024 to February 2023, 2022, etc?
  4. In world sales are making up for the poor marketplace sales performance for me. But yeah, my SLMP income is some of the lowest I think I've ever seen for the last month. My sales have been kind of below normal for several months, but when SLMP really dives the in world sales fill in the gap. If you are selling avatar stuff, the sale events are probably the driving factor. I went as far as to make my own search engine website for my store and it's picking up the slack. But I think there is some fatigue with how difficult it is to find stuff to buy in SL and people aren't bother with SLMP like they used to and instead look for alternate search or events/blogs. I think the last update they did to search greatly increased relevancy of returned results at the cost of the quality and age of listings. The end result was a bunch of very relevant stuff that is horribly outdated. One of the keywords I target has only the first 4 results having ratings, and one is 1 star, the other is 1.5 stars. And further down in results are some 4 and 5 star rated products, some not from my store. In fact some of my lower rated stuff appears higher than my higher rated stuff when searching with two very generic keywords that target both. The new algorithm is also extremely easy to game. I think you have serious merchants not diving into manipulating search in hopes that it'll be fixed, and more scam-tier merchants diving right into gaming the system as hard as they can. I honestly think the days of just depending on SLMP for sales is over.
  5. It's all a mess. I stopped selling full perm for the most part. Even if someone violates the ToS you have a really hard time finding out if they are violating it. Let alone enforcing it. I think it's really just trust, merchants trust customers to follow the ToS and customers respect it so full perm doesn't go away. But it always seems like whenever I release a FP version eventually the product just stops selling and most of my products don't. I would be surprised if most people followed user ToS, let alone read them.
  6. Best way to get high FPS is to lower draw distance, and in crowded places, do things to render fewer avatars, like decrease the complexity limit, increase number of imposters, etc.
  7. When they tried to solve the problem of PBR looking the same in SL and GLTF tools, they managed to really throw a wrench into how users change their environment settings. Sounded good in theory when they were trying to make GLTF tools like Blender and SP have consistency with SL. But it's clear in doing so they completely messed up EEP and how things look for users. I feel like LL expected users to come up with good EEP settings to fill in the void, like SL users usually do. But all the old EEP settings should be moved to "legacy" to make it clear they aren't going to be plug and play with PBR. Which was a really big mistake because nothing angers people like an update that they don't think they need that manages to make them jump through hoops to get things working the way they were. Am I wrong to think that these dark room issues are because of the auto-generated reflection probes at ground level? I haven't played too much but it feels like if you make an enclosed space in the sky you don't end up with dark rooms. Wouldn't it have been better to skip the ground level reflection probes and make it an upgrade, i.e. you buy a house or whatever with probes and lighting set up? With the auto generated probes it feels more like people are getting reflection probes without any lights and it's completely breaking everything, when you need both to get proper results.
  8. Please re-read. It's a lot more difficult to mess up PBR textures than it is spec norm. The problem isn't the PBR textures, it's the environment and scene, i.e. lighting and environment settings. PBR is supposed to emulate real life as much as possible. Please tell me where you can step outside in RL and see a massive blue glow and haze on everything on a sunny day.
  9. There are only a few projector lights rendered at a time. They are only meant to be used sparingly. That said the biggest problem with PBR isn't the PBR content, it's the scene they are in (EEP, reflection probes, etc). It also seems like people don't like the general idea of PBR. PBR is supposed to mimick the real world. A PBR club with proper reflection probes and lighting would be one of the best use cases for PBR. One of the worst use cases is making avatar clothing that is viewed in all sorts of scenes and environments, some which won't be set up for PBR, some will be set up for PBR incorrectly (like the blue glow problem), and some will be set up properly. Which is why people are (understandably) upset about PBR. And about lower end hardware, the default PBR recommended settings are way too high and it's causing a lot of negative feedback. The default, new user PBR experience is horrible between the poor environment settings and the recommended settings being way too high.
  10. 2k/4k would be really good if used properly. If people packed things properly and added multiple materials into a single texture, besides the local improvements, that's also less network requests. Network requests are stuck behind your ping, that's one of the biggest drivers of slow rezzing. A 2k is four times as large as a 1k. Blender's new UV packer works amazingly well and if you are using it, you have to really screw up to make poor use of your space. That said it would completely get abused by people. I imagine clothes with 2k textures on everything just because it looks slightly better than 1k. LL should start a program where they test you on your content creating abilities and it unlocks more advanced features like 2k/4k textures.
  11. I think it entirely depends on if you are limited by your CPU or your GPU with performance. PBR is going to use your GPU more effectively, so if the GPU is starving for data and not being used entirely then PBR is going to be an improvement. If your GPU is already struggling with old viewers, PBR is going to be worse since it's doing so much more stuff. Just my theory at the moment.
  12. I noticed default PBR settings are too aggressive. I.E. what's recommended is usually too high and runs poorly. I was able to get PBR running on a very old netbook. It wouldn't even run at default above 1fps at recommended settings, but lowering draw distance and some settings made it usable if I really had no other choice. What is happening is people are not changing their settings, running PBR, seeing it's slower, then comparing it to their old settings (which probably have been tweaked, like lower draw distance) and thinking PBR is slower. I think Linden Lab is still adjusting their graphical defaults based on their minimum requirements, which seem completely too low https://secondlife.com/system-requirements Do you know anyone running SL with 4GB of ram on an 800mhz Pentium 3? The netbook I used is close to a 1.5ghz Core 2 Quad in performance per clock and it struggled massively.
  13. I'm just assuming you only want one piece of software. If I had to choose I'd take SP over Zbrush but if you can do both you probably should. I'm comparing them based on your workflow, even though each piece of software accomplishes different goals. I think from delivering a quality product, high quality painted materials are more valuable than baking normals to low poly. But obviously the best choice is to do both, I think it just depends how much time you have and what market segment you're targeting. It also depends what you're making. SL used to compress normal maps heavily before PBR, so odds are baking from high to low wouldn't look that great. EllieAnne doesn't have a lot of time and is just getting into things, so I'm basing my opinion on that.
  14. I would take SP over ZBrush because SP makes it a lot easier to handle PBR and GLTF and ZBrush, while superior to Blender in sculpting, only kind of repeats Blender's functionality. Better to ditch ZBrush, learn to sculpt in blender, and use substance painter. AFAIK Blender's GLTF support isn't nearly as good as SP's.
  15. The kit is 8GB, to get that 8GB it's 2 sticks of 4GB. So it's 2 kits * 2 sticks per kit * 4GB per stick = 16. Your math is right, you are just confused by the part that says 8GB. You have two individual kits that are 8GB total, and each kit is 2 sticks of memory. So 4 pieces of memory total that are 4GB each.
  16. Yes, the setting you're looking for is "Brightness (exposure)." It's in the graphics settings. It's a pretty common problem. If you have too much blue in the environment, that can be fixed with the sky settings.
  17. The default environment settings are not good. But you will find the PBR version should use your graphics card a lot better (you should see better FPS if you have a good CPU and graphics card, it's not guaranteed). Nothing wrong with going back to the old way but I suggest you try playing with some settings with your viewer lighting and seeing if you can get something you like.
  18. Surely between this and SLMP having issues they are losing a ton of money. I thought their plan was to shift revenue from land to the SL economy to lower land costs? Clearly this isn't going to work if both are broken most of the time. It's just a matter of time until merchants realize the largest hurdle to making L$ is SL's reliability. If everyone keeps holding L$ because the lindex is broken, everyone is gonna sell at once and the value of L$ is gonna drop quite a bit, at least for a while. Better hope there's enough buyers to be ready for the influx of sales from people waiting to sell L$.
  19. I think this is a symptom of work from home and how it doesn't really work in the long term.
  20. The first upgrade will be to avatars. PBR will be removed and the new lighting system pictured will be deployed. It offers up to 25% better frame rates on lower end hardware.
  21. The number of results in Google are faked. You can not go back that many pages and it ends. 1. Search for Second Life, says 8.5 billion results 2. Turn off javascript 3. Edit the "start" param in the url to 1000 4. 1000 * 10 per page, 10,000 results, right? Should be tons left to reach 8.5 billion 5. No results I don't want to get too off topic but any chance to talk smack about Google, I won't pass up.
  22. I strongly disagree. SLMP is the only real place to buy content for SL besides events. SLMP and XStreet basically killed in world shopping and stores and it's only now kind of coming back because people (merchants and buyers) are upset with reliability and search quality. Unless you have some sort of source to back up your claims, but I always assumed land was the biggest earner followed by people buying and selling L$ to buy stuff on SLMP. Not saying Google is good, I don't use it anymore. But it has the capability to not show you outdated SEO blogspam from 10 years ago when searching. But Google does look at far more than just the content of a web page to rank articles.
  23. Most of the results in Google are all SEO optimized garbage. Most of them follow a very specific formula, being a huge article with a common query in a headline tag followed by a few paragraphs of keyword spam with a sentence or two of relevant information. I think what's concerning is that Google is always changing their algorithm and SEO is a constant battle of trying to figure out how the algorithm has changed. The algo for SLMP search is already known, spam your keyword as much as you can and don't say more than you have to. Google and web is about browser intent, looking for information. SLMP, Amazon, etc are about purchase intent, trying to find something to buy. It's a lot easier to look at some text and figure out what the text is about, than it is to have text about a product and figure out if the product is good enough to rank highly or not. Amazon's primary ranking algorithm is sales velocity (selling faster and faster, or slower or slower, or not at all). It also tracks things like if a visitor to the page bought the product or not. SLMP doesn't do anything like that beyond the top 50k in sales. It's just keyword spam below that. Just my two cents but if Amazon depended entirely on the product descriptions being optimized and had nothing to do with the quality of the product, it wouldn't be very good. I'm not asking SLMP to be as good as Amazon search. But there's so much more it does to determine a quality search result than just looking at text document relevancy. The problem with SLMP is that the search algorithm is so simple that it's extremely easy to game and nothing to hold it back. Someone could easily just sell a prim cube and write up the description properly and appear in the top results for just about anything not major. https://www.anscommerce.com/blog/amazon-seo-strategy/
×
×
  • Create New...